id	sid	tid	token	lemma	pos
ap-990	1	1	ap07_4-5.vp	ap07_4-5.vp	INTJ
ap-990	1	2	1	1	NUM
ap-990	1	3	introduction	introduction	NOUN
ap-990	1	4	automatic	automatic	ADJ
ap-990	1	5	speech	speech	NOUN
ap-990	1	6	recognition	recognition	NOUN
ap-990	1	7	has	have	AUX
ap-990	1	8	become	become	VERB
ap-990	1	9	a	a	DET
ap-990	1	10	very	very	ADV
ap-990	1	11	popular	popular	ADJ
ap-990	1	12	field	field	NOUN
ap-990	1	13	of	of	ADP
ap-990	1	14	research	research	NOUN
ap-990	1	15	,	,	PUNCT
ap-990	1	16	and	and	CCONJ
ap-990	1	17	the	the	DET
ap-990	1	18	results	result	NOUN
ap-990	1	19	come	come	VERB
ap-990	1	20	into	into	ADP
ap-990	1	21	our	our	PRON
ap-990	1	22	lives	life	NOUN
ap-990	1	23	in	in	ADP
ap-990	1	24	many	many	ADJ
ap-990	1	25	forms	form	NOUN
ap-990	1	26	,	,	PUNCT
ap-990	1	27	e.g.	e.g.	ADV
ap-990	1	28	in	in	ADP
ap-990	1	29	voice	voice	NOUN
ap-990	1	30	controlled	control	VERB
ap-990	1	31	machines	machine	NOUN
ap-990	1	32	,	,	PUNCT
ap-990	1	33	like	like	ADP
ap-990	1	34	pcs	pc	NOUN
ap-990	1	35	and	and	CCONJ
ap-990	1	36	mobile	mobile	ADJ
ap-990	1	37	phones	phone	NOUN
ap-990	1	38	.	.	PUNCT
ap-990	2	1	these	these	DET
ap-990	2	2	command	command	NOUN
ap-990	2	3	controlled	control	VERB
ap-990	2	4	systems	system	NOUN
ap-990	2	5	often	often	ADV
ap-990	2	6	work	work	VERB
ap-990	2	7	with	with	ADP
ap-990	2	8	real	real	ADJ
ap-990	2	9	,	,	PUNCT
ap-990	2	10	spontaneous	spontaneous	ADJ
ap-990	2	11	speech	speech	NOUN
ap-990	2	12	,	,	PUNCT
ap-990	2	13	which	which	PRON
ap-990	2	14	is	be	AUX
ap-990	2	15	different	different	ADJ
ap-990	2	16	from	from	ADP
ap-990	2	17	the	the	DET
ap-990	2	18	read	read	NOUN
ap-990	2	19	speech	speech	NOUN
ap-990	2	20	used	use	VERB
ap-990	2	21	in	in	ADP
ap-990	2	22	the	the	DET
ap-990	2	23	clean	clean	ADJ
ap-990	2	24	laboratory	laboratory	NOUN
ap-990	2	25	conditions	condition	NOUN
ap-990	2	26	of	of	ADP
ap-990	2	27	a	a	DET
ap-990	2	28	research	research	NOUN
ap-990	2	29	centre	centre	NOUN
ap-990	2	30	.	.	PUNCT
ap-990	3	1	this	this	DET
ap-990	3	2	noisy	noisy	ADJ
ap-990	3	3	kind	kind	NOUN
ap-990	3	4	of	of	ADP
ap-990	3	5	speech	speech	NOUN
ap-990	3	6	is	be	AUX
ap-990	3	7	recorded	record	VERB
ap-990	3	8	in	in	ADP
ap-990	3	9	a	a	DET
ap-990	3	10	real	real	ADJ
ap-990	3	11	environment	environment	NOUN
ap-990	3	12	,	,	PUNCT
ap-990	3	13	which	which	PRON
ap-990	3	14	causes	cause	VERB
ap-990	3	15	some	some	DET
ap-990	3	16	noise	noise	NOUN
ap-990	3	17	addition	addition	NOUN
ap-990	3	18	to	to	ADP
ap-990	3	19	the	the	DET
ap-990	3	20	speech	speech	NOUN
ap-990	3	21	signal	signal	NOUN
ap-990	3	22	.	.	PUNCT
ap-990	4	1	the	the	DET
ap-990	4	2	speech	speech	NOUN
ap-990	4	3	is	be	AUX
ap-990	4	4	also	also	ADV
ap-990	4	5	created	create	VERB
ap-990	4	6	”	"	PUNCT
ap-990	4	7	on	on	ADP
ap-990	4	8	the	the	DET
ap-990	4	9	fly	fly	NOUN
ap-990	4	10	”	"	PUNCT
ap-990	4	11	and	and	CCONJ
ap-990	4	12	the	the	DET
ap-990	4	13	speaker	speaker	NOUN
ap-990	4	14	has	have	VERB
ap-990	4	15	to	to	PART
ap-990	4	16	think	think	VERB
ap-990	4	17	of	of	ADP
ap-990	4	18	the	the	DET
ap-990	4	19	next	next	ADJ
ap-990	4	20	word	word	NOUN
ap-990	4	21	while	while	SCONJ
ap-990	4	22	speaking	speak	VERB
ap-990	4	23	.	.	PUNCT
ap-990	5	1	all	all	DET
ap-990	5	2	this	this	PRON
ap-990	5	3	causes	cause	VERB
ap-990	5	4	many	many	ADJ
ap-990	5	5	non	non	ADJ
ap-990	5	6	-	-	ADJ
ap-990	5	7	speech	speech	ADJ
ap-990	5	8	events	event	NOUN
ap-990	5	9	to	to	PART
ap-990	5	10	be	be	AUX
ap-990	5	11	present	present	ADJ
ap-990	5	12	in	in	ADP
ap-990	5	13	such	such	ADJ
ap-990	5	14	speech	speech	NOUN
ap-990	5	15	.	.	PUNCT
ap-990	6	1	therefore	therefore	ADV
ap-990	6	2	it	it	PRON
ap-990	6	3	is	be	AUX
ap-990	6	4	important	important	ADJ
ap-990	6	5	to	to	PART
ap-990	6	6	make	make	VERB
ap-990	6	7	the	the	DET
ap-990	6	8	recogniser	recogniser	NOUN
ap-990	6	9	robust	robust	ADJ
ap-990	6	10	against	against	ADP
ap-990	6	11	these	these	DET
ap-990	6	12	events	event	NOUN
ap-990	6	13	,	,	PUNCT
ap-990	6	14	and	and	CCONJ
ap-990	6	15	to	to	PART
ap-990	6	16	take	take	VERB
ap-990	6	17	the	the	DET
ap-990	6	18	events	event	NOUN
ap-990	6	19	into	into	ADP
ap-990	6	20	account	account	NOUN
ap-990	6	21	and	and	CCONJ
ap-990	6	22	ensure	ensure	VERB
ap-990	6	23	that	that	SCONJ
ap-990	6	24	they	they	PRON
ap-990	6	25	are	be	AUX
ap-990	6	26	not	not	PART
ap-990	6	27	incorrectly	incorrectly	ADV
ap-990	6	28	recognised	recognise	VERB
ap-990	6	29	.	.	PUNCT
ap-990	7	1	the	the	DET
ap-990	7	2	long	long	ADJ
ap-990	7	3	-	-	PUNCT
ap-990	7	4	term	term	NOUN
ap-990	7	5	goal	goal	NOUN
ap-990	7	6	of	of	ADP
ap-990	7	7	our	our	PRON
ap-990	7	8	work	work	NOUN
ap-990	7	9	is	be	AUX
ap-990	7	10	to	to	PART
ap-990	7	11	create	create	VERB
ap-990	7	12	a	a	DET
ap-990	7	13	hidden	hide	VERB
ap-990	7	14	markov	markov	NOUN
ap-990	7	15	model	model	NOUN
ap-990	7	16	(	(	PUNCT
ap-990	7	17	hmm	hmm	INTJ
ap-990	7	18	)	)	PUNCT
ap-990	7	19	based	base	VERB
ap-990	7	20	digit	digit	NOUN
ap-990	7	21	recogniser	recogniser	PROPN
ap-990	7	22	,	,	PUNCT
ap-990	7	23	which	which	PRON
ap-990	7	24	will	will	AUX
ap-990	7	25	suppress	suppress	VERB
ap-990	7	26	the	the	DET
ap-990	7	27	non	non	ADJ
ap-990	7	28	-	-	ADJ
ap-990	7	29	speech	speech	ADJ
ap-990	7	30	event	event	NOUN
ap-990	7	31	influence	influence	NOUN
ap-990	7	32	.	.	PUNCT
ap-990	8	1	one	one	NUM
ap-990	8	2	solution	solution	NOUN
ap-990	8	3	is	be	AUX
ap-990	8	4	to	to	PART
ap-990	8	5	model	model	VERB
ap-990	8	6	the	the	DET
ap-990	8	7	non	non	ADJ
ap-990	8	8	-	-	ADJ
ap-990	8	9	speech	speech	ADJ
ap-990	8	10	events	event	NOUN
ap-990	8	11	that	that	PRON
ap-990	8	12	affect	affect	VERB
ap-990	8	13	speech	speech	NOUN
ap-990	8	14	.	.	PUNCT
ap-990	9	1	for	for	ADP
ap-990	9	2	this	this	DET
ap-990	9	3	purpose	purpose	NOUN
ap-990	9	4	it	it	PRON
ap-990	9	5	is	be	AUX
ap-990	9	6	important	important	ADJ
ap-990	9	7	to	to	PART
ap-990	9	8	have	have	VERB
ap-990	9	9	a	a	DET
ap-990	9	10	large	large	ADJ
ap-990	9	11	training	training	NOUN
ap-990	9	12	database	database	NOUN
ap-990	9	13	with	with	ADP
ap-990	9	14	many	many	ADJ
ap-990	9	15	occurrences	occurrence	NOUN
ap-990	9	16	of	of	ADP
ap-990	9	17	each	each	DET
ap-990	9	18	modelled	model	VERB
ap-990	9	19	item	item	NOUN
ap-990	9	20	,	,	PUNCT
ap-990	9	21	e.g.	e.g.	ADV
ap-990	9	22	phoneme	phoneme	NOUN
ap-990	9	23	,	,	PUNCT
ap-990	9	24	to	to	PART
ap-990	9	25	get	get	VERB
ap-990	9	26	a	a	DET
ap-990	9	27	general	general	ADJ
ap-990	9	28	description	description	NOUN
ap-990	9	29	of	of	ADP
ap-990	9	30	the	the	DET
ap-990	9	31	item	item	NOUN
ap-990	9	32	.	.	PUNCT
ap-990	10	1	therefore	therefore	ADV
ap-990	10	2	in	in	ADP
ap-990	10	3	the	the	DET
ap-990	10	4	first	first	ADJ
ap-990	10	5	part	part	NOUN
ap-990	10	6	of	of	ADP
ap-990	10	7	this	this	DET
ap-990	10	8	work	work	NOUN
ap-990	10	9	,	,	PUNCT
ap-990	10	10	two	two	NUM
ap-990	10	11	czech	czech	ADJ
ap-990	10	12	speech	speech	NOUN
ap-990	10	13	databases	database	NOUN
ap-990	10	14	are	be	AUX
ap-990	10	15	analysed	analyse	VERB
ap-990	10	16	for	for	ADP
ap-990	10	17	the	the	DET
ap-990	10	18	presence	presence	NOUN
ap-990	10	19	and	and	CCONJ
ap-990	10	20	quality	quality	NOUN
ap-990	10	21	of	of	ADP
ap-990	10	22	the	the	DET
ap-990	10	23	speaker	speaker	NOUN
ap-990	10	24	non	non	ADJ
ap-990	10	25	-	-	ADJ
ap-990	10	26	speech	speech	ADJ
ap-990	10	27	events	event	NOUN
ap-990	10	28	.	.	PUNCT
ap-990	11	1	in	in	ADP
ap-990	11	2	this	this	DET
ap-990	11	3	very	very	ADJ
ap-990	11	4	beginning	beginning	NOUN
ap-990	11	5	of	of	ADP
ap-990	11	6	our	our	PRON
ap-990	11	7	work	work	NOUN
ap-990	11	8	,	,	PUNCT
ap-990	11	9	only	only	ADV
ap-990	11	10	speaker	speaker	NOUN
ap-990	11	11	-	-	PUNCT
ap-990	11	12	generated	generate	VERB
ap-990	11	13	non	non	ADJ
ap-990	11	14	-	-	ADJ
ap-990	11	15	speech	speech	ADJ
ap-990	11	16	events	event	NOUN
ap-990	11	17	are	be	AUX
ap-990	11	18	taken	take	VERB
ap-990	11	19	in	in	ADP
ap-990	11	20	account	account	NOUN
ap-990	11	21	because	because	SCONJ
ap-990	11	22	of	of	ADP
ap-990	11	23	their	their	PRON
ap-990	11	24	position	position	NOUN
ap-990	11	25	between	between	ADP
ap-990	11	26	words	word	NOUN
ap-990	11	27	of	of	ADP
ap-990	11	28	recognised	recognise	VERB
ap-990	11	29	speech	speech	NOUN
ap-990	11	30	.	.	PUNCT
ap-990	12	1	this	this	PRON
ap-990	12	2	enables	enable	VERB
ap-990	12	3	us	we	PRON
ap-990	12	4	to	to	PART
ap-990	12	5	consider	consider	VERB
ap-990	12	6	the	the	DET
ap-990	12	7	events	event	NOUN
ap-990	12	8	to	to	PART
ap-990	12	9	be	be	AUX
ap-990	12	10	another	another	DET
ap-990	12	11	word	word	NOUN
ap-990	12	12	and	and	CCONJ
ap-990	12	13	to	to	PART
ap-990	12	14	model	model	VERB
ap-990	12	15	them	they	PRON
ap-990	12	16	easily	easily	ADV
ap-990	12	17	.	.	PUNCT
ap-990	13	1	for	for	ADP
ap-990	13	2	this	this	DET
ap-990	13	3	purpose	purpose	NOUN
ap-990	13	4	,	,	PUNCT
ap-990	13	5	the	the	DET
ap-990	13	6	databases	database	NOUN
ap-990	13	7	were	be	AUX
ap-990	13	8	inspected	inspect	VERB
ap-990	13	9	for	for	ADP
ap-990	13	10	the	the	DET
ap-990	13	11	presence	presence	NOUN
ap-990	13	12	of	of	ADP
ap-990	13	13	the	the	DET
ap-990	13	14	events	event	NOUN
ap-990	13	15	in	in	ADP
ap-990	13	16	different	different	ADJ
ap-990	13	17	situations	situation	NOUN
ap-990	13	18	.	.	PUNCT
ap-990	14	1	the	the	DET
ap-990	14	2	second	second	ADJ
ap-990	14	3	part	part	NOUN
ap-990	14	4	of	of	ADP
ap-990	14	5	the	the	DET
ap-990	14	6	work	work	NOUN
ap-990	14	7	is	be	AUX
ap-990	14	8	concerned	concern	VERB
ap-990	14	9	with	with	ADP
ap-990	14	10	speaker	speaker	NOUN
ap-990	14	11	non	non	ADJ
ap-990	14	12	-	-	ADJ
ap-990	14	13	speech	speech	ADJ
ap-990	14	14	event	event	NOUN
ap-990	14	15	modelling	modelling	NOUN
ap-990	14	16	.	.	PUNCT
ap-990	15	1	a	a	DET
ap-990	15	2	robust	robust	ADJ
ap-990	15	3	czech	czech	ADJ
ap-990	15	4	digit	digit	NOUN
ap-990	15	5	sequence	sequence	NOUN
ap-990	15	6	recogniser	recogniser	NOUN
ap-990	15	7	based	base	VERB
ap-990	15	8	on	on	ADP
ap-990	15	9	hmms	hmms	NOUN
ap-990	15	10	of	of	ADP
ap-990	15	11	czech	czech	ADJ
ap-990	15	12	phonemes	phoneme	NOUN
ap-990	15	13	is	be	AUX
ap-990	15	14	trained	train	VERB
ap-990	15	15	on	on	ADP
ap-990	15	16	the	the	DET
ap-990	15	17	analysed	analyse	VERB
ap-990	15	18	database	database	NOUN
ap-990	15	19	,	,	PUNCT
ap-990	15	20	and	and	CCONJ
ap-990	15	21	the	the	DET
ap-990	15	22	results	result	NOUN
ap-990	15	23	gained	gain	VERB
ap-990	15	24	with	with	ADP
ap-990	15	25	the	the	DET
ap-990	15	26	non	non	ADJ
ap-990	15	27	-	-	ADJ
ap-990	15	28	speech	speech	ADJ
ap-990	15	29	event	event	NOUN
ap-990	15	30	recognition	recognition	NOUN
ap-990	15	31	feature	feature	NOUN
ap-990	15	32	are	be	AUX
ap-990	15	33	presented	present	VERB
ap-990	15	34	.	.	PUNCT
ap-990	16	1	an	an	DET
ap-990	16	2	hmm	hmm	NOUN
ap-990	16	3	toolkit	toolkit	NOUN
ap-990	16	4	[	[	X
ap-990	16	5	2	2	X
ap-990	16	6	]	]	PUNCT
ap-990	16	7	is	be	AUX
ap-990	16	8	used	use	VERB
ap-990	16	9	for	for	ADP
ap-990	16	10	this	this	DET
ap-990	16	11	purpose	purpose	NOUN
ap-990	16	12	.	.	PUNCT
ap-990	17	1	based	base	VERB
ap-990	17	2	on	on	ADP
ap-990	17	3	these	these	DET
ap-990	17	4	results	result	NOUN
ap-990	17	5	,	,	PUNCT
ap-990	17	6	forced	force	VERB
ap-990	17	7	-	-	PUNCT
ap-990	17	8	alignment	alignment	NOUN
ap-990	17	9	experiments	experiment	NOUN
ap-990	17	10	are	be	AUX
ap-990	17	11	shown	show	VERB
ap-990	17	12	in	in	ADP
ap-990	17	13	the	the	DET
ap-990	17	14	third	third	ADJ
ap-990	17	15	part	part	NOUN
ap-990	17	16	of	of	ADP
ap-990	17	17	the	the	DET
ap-990	17	18	work	work	NOUN
ap-990	17	19	.	.	PUNCT
ap-990	18	1	the	the	DET
ap-990	18	2	recogniser	recogniser	PROPN
ap-990	18	3	is	be	AUX
ap-990	18	4	used	use	VERB
ap-990	18	5	to	to	PART
ap-990	18	6	re	re	VERB
ap-990	18	7	-	-	VERB
ap-990	18	8	recognise	recognise	VERB
ap-990	18	9	the	the	DET
ap-990	18	10	training	training	NOUN
ap-990	18	11	data	datum	NOUN
ap-990	18	12	.	.	PUNCT
ap-990	19	1	this	this	PRON
ap-990	19	2	will	will	AUX
ap-990	19	3	remove	remove	VERB
ap-990	19	4	unsuitable	unsuitable	ADJ
ap-990	19	5	event	event	NOUN
ap-990	19	6	marks	mark	NOUN
ap-990	19	7	in	in	ADP
ap-990	19	8	the	the	DET
ap-990	19	9	transcription	transcription	NOUN
ap-990	19	10	and	and	CCONJ
ap-990	19	11	it	it	PRON
ap-990	19	12	also	also	ADV
ap-990	19	13	enables	enable	VERB
ap-990	19	14	to	to	PART
ap-990	19	15	remark	remark	VERB
ap-990	19	16	the	the	DET
ap-990	19	17	event	event	NOUN
ap-990	19	18	,	,	PUNCT
ap-990	19	19	which	which	PRON
ap-990	19	20	should	should	AUX
ap-990	19	21	improve	improve	VERB
ap-990	19	22	the	the	DET
ap-990	19	23	recognition	recognition	NOUN
ap-990	19	24	results	result	NOUN
ap-990	19	25	.	.	PUNCT
ap-990	20	1	2	2	NUM
ap-990	20	2	the	the	DET
ap-990	20	3	database	database	NOUN
ap-990	20	4	for	for	ADP
ap-990	20	5	the	the	DET
ap-990	20	6	non	non	ADJ
ap-990	20	7	-	-	ADJ
ap-990	20	8	speech	speech	ADJ
ap-990	20	9	event	event	NOUN
ap-990	20	10	recognition	recognition	NOUN
ap-990	20	11	task	task	NOUN
ap-990	20	12	as	as	SCONJ
ap-990	20	13	noted	note	VERB
ap-990	20	14	above	above	ADV
ap-990	20	15	,	,	PUNCT
ap-990	20	16	it	it	PRON
ap-990	20	17	is	be	AUX
ap-990	20	18	important	important	ADJ
ap-990	20	19	for	for	SCONJ
ap-990	20	20	automatic	automatic	ADJ
ap-990	20	21	speech	speech	NOUN
ap-990	20	22	recogniser	recogniser	NOUN
ap-990	20	23	training	training	NOUN
ap-990	20	24	to	to	PART
ap-990	20	25	have	have	VERB
ap-990	20	26	a	a	DET
ap-990	20	27	speech	speech	NOUN
ap-990	20	28	database	database	NOUN
ap-990	20	29	that	that	PRON
ap-990	20	30	has	have	VERB
ap-990	20	31	enough	enough	ADJ
ap-990	20	32	occurrences	occurrence	NOUN
ap-990	20	33	of	of	ADP
ap-990	20	34	each	each	DET
ap-990	20	35	type	type	NOUN
ap-990	20	36	of	of	ADP
ap-990	20	37	modelled	model	VERB
ap-990	20	38	unit	unit	NOUN
ap-990	20	39	,	,	PUNCT
ap-990	20	40	e.g.	e.g.	ADV
ap-990	20	41	phonemes	phoneme	NOUN
ap-990	20	42	.	.	PUNCT
ap-990	21	1	in	in	ADP
ap-990	21	2	the	the	DET
ap-990	21	3	non	non	ADJ
ap-990	21	4	-	-	ADJ
ap-990	21	5	speech	speech	ADJ
ap-990	21	6	event	event	NOUN
ap-990	21	7	modelling	model	VERB
ap-990	21	8	task	task	NOUN
ap-990	21	9	this	this	PRON
ap-990	21	10	calls	call	VERB
ap-990	21	11	for	for	ADP
ap-990	21	12	a	a	DET
ap-990	21	13	sufficient	sufficient	ADJ
ap-990	21	14	number	number	NOUN
ap-990	21	15	of	of	ADP
ap-990	21	16	non	non	ADJ
ap-990	21	17	-	-	ADJ
ap-990	21	18	speech	speech	ADJ
ap-990	21	19	events	event	NOUN
ap-990	21	20	present	present	ADJ
ap-990	21	21	in	in	ADP
ap-990	21	22	the	the	DET
ap-990	21	23	database	database	NOUN
ap-990	21	24	,	,	PUNCT
ap-990	21	25	which	which	PRON
ap-990	21	26	is	be	AUX
ap-990	21	27	the	the	DET
ap-990	21	28	way	way	NOUN
ap-990	21	29	to	to	PART
ap-990	21	30	get	get	VERB
ap-990	21	31	a	a	DET
ap-990	21	32	general	general	ADJ
ap-990	21	33	description	description	NOUN
ap-990	21	34	of	of	ADP
ap-990	21	35	the	the	DET
ap-990	21	36	modelled	model	VERB
ap-990	21	37	event	event	NOUN
ap-990	21	38	.	.	PUNCT
ap-990	22	1	this	this	DET
ap-990	22	2	work	work	NOUN
ap-990	22	3	deals	deal	VERB
ap-990	22	4	with	with	ADP
ap-990	22	5	speaker	speaker	NOUN
ap-990	22	6	non	non	ADJ
ap-990	22	7	-	-	NOUN
ap-990	22	8	speech	speech	ADJ
ap-990	22	9	events	event	NOUN
ap-990	22	10	that	that	PRON
ap-990	22	11	appear	appear	VERB
ap-990	22	12	in	in	ADP
ap-990	22	13	between	between	ADP
ap-990	22	14	particular	particular	ADJ
ap-990	22	15	words	word	NOUN
ap-990	22	16	,	,	PUNCT
ap-990	22	17	so	so	SCONJ
ap-990	22	18	the	the	DET
ap-990	22	19	database	database	NOUN
ap-990	22	20	is	be	AUX
ap-990	22	21	analysed	analyse	VERB
ap-990	22	22	for	for	ADP
ap-990	22	23	the	the	DET
ap-990	22	24	presence	presence	NOUN
ap-990	22	25	of	of	ADP
ap-990	22	26	these	these	DET
ap-990	22	27	events	event	NOUN
ap-990	22	28	.	.	PUNCT
ap-990	23	1	in	in	ADP
ap-990	23	2	most	most	ADJ
ap-990	23	3	cases	case	NOUN
ap-990	23	4	,	,	PUNCT
ap-990	23	5	speech	speech	NOUN
ap-990	23	6	recognition	recognition	NOUN
ap-990	23	7	systems	system	NOUN
ap-990	23	8	are	be	AUX
ap-990	23	9	trained	train	VERB
ap-990	23	10	on	on	ADP
ap-990	23	11	a	a	DET
ap-990	23	12	database	database	NOUN
ap-990	23	13	of	of	ADP
ap-990	23	14	mainly	mainly	ADV
ap-990	23	15	read	read	VERB
ap-990	23	16	speech	speech	NOUN
ap-990	23	17	.	.	PUNCT
ap-990	24	1	this	this	PRON
ap-990	24	2	helps	help	VERB
ap-990	24	3	to	to	PART
ap-990	24	4	complete	complete	VERB
ap-990	24	5	the	the	DET
ap-990	24	6	database	database	NOUN
ap-990	24	7	,	,	PUNCT
ap-990	24	8	because	because	SCONJ
ap-990	24	9	it	it	PRON
ap-990	24	10	is	be	AUX
ap-990	24	11	known	know	VERB
ap-990	24	12	approximately	approximately	ADV
ap-990	24	13	what	what	PRON
ap-990	24	14	was	be	AUX
ap-990	24	15	said	say	VERB
ap-990	24	16	.	.	PUNCT
ap-990	25	1	however	however	ADV
ap-990	25	2	,	,	PUNCT
ap-990	25	3	read	read	VERB
ap-990	25	4	speech	speech	NOUN
ap-990	25	5	is	be	AUX
ap-990	25	6	different	different	ADJ
ap-990	25	7	from	from	ADP
ap-990	25	8	the	the	DET
ap-990	25	9	spontaneous	spontaneous	ADJ
ap-990	25	10	speech	speech	NOUN
ap-990	25	11	used	use	VERB
ap-990	25	12	in	in	ADP
ap-990	25	13	voice	voice	NOUN
ap-990	25	14	communication	communication	NOUN
ap-990	25	15	with	with	ADP
ap-990	25	16	a	a	DET
ap-990	25	17	machine	machine	NOUN
ap-990	25	18	.	.	PUNCT
ap-990	26	1	read	read	NOUN
ap-990	26	2	speech	speech	NOUN
ap-990	26	3	is	be	AUX
ap-990	26	4	low	low	ADJ
ap-990	26	5	in	in	ADP
ap-990	26	6	speaker	speaker	NOUN
ap-990	26	7	non	non	ADJ
ap-990	26	8	-	-	ADJ
ap-990	26	9	speech	speech	ADJ
ap-990	26	10	events	event	NOUN
ap-990	26	11	.	.	PUNCT
ap-990	27	1	unlike	unlike	ADP
ap-990	27	2	a	a	DET
ap-990	27	3	spontaneous	spontaneous	ADJ
ap-990	27	4	speaker	speaker	NOUN
ap-990	27	5	,	,	PUNCT
ap-990	27	6	a	a	DET
ap-990	27	7	reader	reader	NOUN
ap-990	27	8	is	be	AUX
ap-990	27	9	not	not	PART
ap-990	27	10	forced	force	VERB
ap-990	27	11	to	to	PART
ap-990	27	12	think	think	VERB
ap-990	27	13	while	while	SCONJ
ap-990	27	14	speaking	speak	VERB
ap-990	27	15	and	and	CCONJ
ap-990	27	16	so	so	ADV
ap-990	27	17	the	the	DET
ap-990	27	18	occurrence	occurrence	NOUN
ap-990	27	19	of	of	ADP
ap-990	27	20	hesitation	hesitation	NOUN
ap-990	27	21	is	be	AUX
ap-990	27	22	low	low	ADJ
ap-990	27	23	.	.	PUNCT
ap-990	28	1	the	the	DET
ap-990	28	2	same	same	ADJ
ap-990	28	3	problem	problem	NOUN
ap-990	28	4	occurs	occur	VERB
ap-990	28	5	with	with	ADP
ap-990	28	6	other	other	ADJ
ap-990	28	7	possible	possible	ADJ
ap-990	28	8	non	non	ADJ
ap-990	28	9	-	-	ADJ
ap-990	28	10	speech	speech	ADJ
ap-990	28	11	events	event	NOUN
ap-990	28	12	,	,	PUNCT
ap-990	28	13	like	like	ADP
ap-990	28	14	lip	lip	NOUN
ap-990	28	15	-	-	PUNCT
ap-990	28	16	smacking	smacking	NOUN
ap-990	28	17	,	,	PUNCT
ap-990	28	18	because	because	SCONJ
ap-990	28	19	the	the	DET
ap-990	28	20	reader	reader	NOUN
ap-990	28	21	has	have	VERB
ap-990	28	22	better	well	ADJ
ap-990	28	23	control	control	NOUN
ap-990	28	24	over	over	ADP
ap-990	28	25	his	his	PRON
ap-990	28	26	mouth	mouth	NOUN
ap-990	28	27	.	.	PUNCT
ap-990	29	1	the	the	DET
ap-990	29	2	recognizer	recognizer	NOUN
ap-990	29	3	presented	present	VERB
ap-990	29	4	here	here	ADV
ap-990	29	5	was	be	AUX
ap-990	29	6	trained	train	VERB
ap-990	29	7	on	on	ADP
ap-990	29	8	two	two	NUM
ap-990	29	9	different	different	ADJ
ap-990	29	10	sets	set	NOUN
ap-990	29	11	of	of	ADP
ap-990	29	12	speech	speech	NOUN
ap-990	29	13	.	.	PUNCT
ap-990	30	1	the	the	DET
ap-990	30	2	first	first	ADJ
ap-990	30	3	database	database	NOUN
ap-990	30	4	(	(	PUNCT
ap-990	30	5	spee	spee	NOUN
ap-990	30	6	)	)	PUNCT
ap-990	30	7	is	be	AUX
ap-990	30	8	a	a	DET
ap-990	30	9	collection	collection	NOUN
ap-990	30	10	of	of	ADP
ap-990	30	11	czech	czech	ADJ
ap-990	30	12	speaker	speaker	NOUN
ap-990	30	13	records	record	NOUN
ap-990	30	14	in	in	ADP
ap-990	30	15	a	a	DET
ap-990	30	16	different	different	ADJ
ap-990	30	17	environment	environment	NOUN
ap-990	30	18	,	,	PUNCT
ap-990	30	19	which	which	PRON
ap-990	30	20	was	be	AUX
ap-990	30	21	inspected	inspect	VERB
ap-990	30	22	to	to	PART
ap-990	30	23	contain	contain	VERB
ap-990	30	24	only	only	ADV
ap-990	30	25	the	the	DET
ap-990	30	26	items	item	NOUN
ap-990	30	27	in	in	ADP
ap-990	30	28	a	a	DET
ap-990	30	29	silent	silent	ADJ
ap-990	30	30	environment	environment	NOUN
ap-990	30	31	for	for	ADP
ap-990	30	32	training	training	NOUN
ap-990	30	33	purposes	purpose	NOUN
ap-990	30	34	.	.	PUNCT
ap-990	31	1	the	the	DET
ap-990	31	2	spee	spee	NOUN
ap-990	31	3	set	set	NOUN
ap-990	31	4	includes	include	VERB
ap-990	31	5	speaker	speaker	NOUN
ap-990	31	6	non	non	ADJ
ap-990	31	7	-	-	ADJ
ap-990	31	8	speech	speech	ADJ
ap-990	31	9	events	event	NOUN
ap-990	31	10	divided	divide	VERB
ap-990	31	11	into	into	ADP
ap-990	31	12	two	two	NUM
ap-990	31	13	classes	class	NOUN
ap-990	31	14	:	:	PUNCT
ap-990	31	15	filled	fill	VERB
ap-990	31	16	pauses	pause	NOUN
ap-990	31	17	(	(	PUNCT
ap-990	31	18	fil	fil	NOUN
ap-990	31	19	)	)	PUNCT
ap-990	31	20	,	,	PUNCT
ap-990	31	21	which	which	PRON
ap-990	31	22	are	be	AUX
ap-990	31	23	pauses	pause	NOUN
ap-990	31	24	in	in	ADP
ap-990	31	25	speech	speech	NOUN
ap-990	31	26	filled	fill	VERB
ap-990	31	27	with	with	ADP
ap-990	31	28	some	some	DET
ap-990	31	29	sound	sound	NOUN
ap-990	31	30	(	(	PUNCT
ap-990	31	31	common	common	ADJ
ap-990	31	32	in	in	ADP
ap-990	31	33	hesitation	hesitation	NOUN
ap-990	31	34	)	)	PUNCT
ap-990	31	35	,	,	PUNCT
ap-990	31	36	and	and	CCONJ
ap-990	31	37	other	other	ADJ
ap-990	31	38	speaker	speaker	NOUN
ap-990	31	39	non	non	ADJ
ap-990	31	40	-	-	ADJ
ap-990	31	41	speech	speech	ADJ
ap-990	31	42	events	event	NOUN
ap-990	31	43	(	(	PUNCT
ap-990	31	44	spk	spk	PROPN
ap-990	31	45	)	)	PUNCT
ap-990	31	46	.	.	PUNCT
ap-990	32	1	to	to	PART
ap-990	32	2	increase	increase	VERB
ap-990	32	3	the	the	DET
ap-990	32	4	number	number	NOUN
ap-990	32	5	of	of	ADP
ap-990	32	6	non	non	ADJ
ap-990	32	7	-	-	ADJ
ap-990	32	8	speech	speech	ADJ
ap-990	32	9	events	event	NOUN
ap-990	32	10	present	present	ADJ
ap-990	32	11	in	in	ADP
ap-990	32	12	the	the	DET
ap-990	32	13	training	training	NOUN
ap-990	32	14	set	set	NOUN
ap-990	32	15	of	of	ADP
ap-990	32	16	speech	speech	NOUN
ap-990	32	17	data	datum	NOUN
ap-990	32	18	,	,	PUNCT
ap-990	32	19	one	one	NUM
ap-990	32	20	other	other	ADJ
ap-990	32	21	dataset	dataset	NOUN
ap-990	32	22	was	be	AUX
ap-990	32	23	used	use	VERB
ap-990	32	24	.	.	PUNCT
ap-990	33	1	it	it	PRON
ap-990	33	2	is	be	AUX
ap-990	33	3	a	a	DET
ap-990	33	4	collection	collection	NOUN
ap-990	33	5	of	of	ADP
ap-990	33	6	records	record	NOUN
ap-990	33	7	from	from	ADP
ap-990	33	8	a	a	DET
ap-990	33	9	car	car	NOUN
ap-990	33	10	(	(	PUNCT
ap-990	33	11	tem	tem	PROPN
ap-990	33	12	)	)	PUNCT
ap-990	33	13	,	,	PUNCT
ap-990	33	14	which	which	PRON
ap-990	33	15	was	be	AUX
ap-990	33	16	also	also	ADV
ap-990	33	17	inspected	inspect	VERB
ap-990	33	18	to	to	PART
ap-990	33	19	contain	contain	VERB
ap-990	33	20	only	only	ADV
ap-990	33	21	silent	silent	ADJ
ap-990	33	22	items	item	NOUN
ap-990	33	23	from	from	ADP
ap-990	33	24	a	a	DET
ap-990	33	25	standing	standing	ADJ
ap-990	33	26	car	car	NOUN
ap-990	33	27	.	.	PUNCT
ap-990	34	1	this	this	DET
ap-990	34	2	database	database	NOUN
ap-990	34	3	divides	divide	VERB
ap-990	34	4	the	the	DET
ap-990	34	5	events	event	NOUN
ap-990	34	6	into	into	ADP
ap-990	34	7	several	several	ADJ
ap-990	34	8	classes	class	NOUN
ap-990	34	9	,	,	PUNCT
ap-990	34	10	which	which	PRON
ap-990	34	11	helps	help	VERB
ap-990	34	12	to	to	PART
ap-990	34	13	create	create	VERB
ap-990	34	14	more	more	ADV
ap-990	34	15	different	different	ADJ
ap-990	34	16	models	model	NOUN
ap-990	34	17	of	of	ADP
ap-990	34	18	non	non	ADJ
ap-990	34	19	-	-	ADJ
ap-990	34	20	speech	speech	ADJ
ap-990	34	21	events	event	NOUN
ap-990	34	22	and	and	CCONJ
ap-990	34	23	to	to	PART
ap-990	34	24	describe	describe	VERB
ap-990	34	25	these	these	DET
ap-990	34	26	events	event	NOUN
ap-990	34	27	more	more	ADV
ap-990	34	28	accurately	accurately	ADV
ap-990	34	29	.	.	PUNCT
ap-990	35	1	table	table	NOUN
ap-990	35	2	1	1	NUM
ap-990	35	3	shows	show	VERB
ap-990	35	4	the	the	DET
ap-990	35	5	number	number	NOUN
ap-990	35	6	of	of	ADP
ap-990	35	7	events	event	NOUN
ap-990	35	8	marked	mark	VERB
ap-990	35	9	in	in	ADP
ap-990	35	10	the	the	DET
ap-990	35	11	transcription	transcription	NOUN
ap-990	35	12	of	of	ADP
ap-990	35	13	the	the	DET
ap-990	35	14	whole	whole	ADJ
ap-990	35	15	dataset	dataset	NOUN
ap-990	35	16	and	and	CCONJ
ap-990	35	17	of	of	ADP
ap-990	35	18	the	the	DET
ap-990	35	19	selected	select	VERB
ap-990	35	20	clean	clean	ADJ
ap-990	35	21	training	training	NOUN
ap-990	35	22	subset	subset	NOUN
ap-990	35	23	for	for	SCONJ
ap-990	35	24	both	both	CCONJ
ap-990	35	25	the	the	DET
ap-990	35	26	spee	spee	NOUN
ap-990	35	27	and	and	CCONJ
ap-990	35	28	the	the	DET
ap-990	35	29	tem	tem	PROPN
ap-990	35	30	dataset	dataset	VERB
ap-990	35	31	.	.	PUNCT
ap-990	36	1	it	it	PRON
ap-990	36	2	can	can	AUX
ap-990	36	3	be	be	AUX
ap-990	36	4	seen	see	VERB
ap-990	36	5	that	that	SCONJ
ap-990	36	6	the	the	DET
ap-990	36	7	training	training	NOUN
ap-990	36	8	selections	selection	NOUN
ap-990	36	9	(	(	PUNCT
ap-990	36	10	clean	clean	ADJ
ap-990	36	11	)	)	PUNCT
ap-990	36	12	contain	contain	VERB
ap-990	36	13	notably	notably	ADV
ap-990	36	14	fewer	few	ADJ
ap-990	36	15	speaker	speaker	NOUN
ap-990	36	16	non	non	ADJ
ap-990	36	17	-	-	ADJ
ap-990	36	18	speech	speech	ADJ
ap-990	36	19	events	event	NOUN
ap-990	36	20	marked	mark	VERB
ap-990	36	21	in	in	ADP
ap-990	36	22	the	the	DET
ap-990	36	23	tran	tran	PROPN
ap-990	36	24	©	©	PROPN
ap-990	36	25	czech	czech	PROPN
ap-990	36	26	technical	technical	PROPN
ap-990	36	27	university	university	PROPN
ap-990	36	28	publishing	publishing	NOUN
ap-990	36	29	house	house	NOUN
ap-990	36	30	http://ctn.cvut.cz/ap/	http://ctn.cvut.cz/ap/	PROPN
ap-990	36	31	107	107	NUM
ap-990	36	32	acta	acta	PROPN
ap-990	36	33	polytechnica	polytechnica	PROPN
ap-990	36	34	vol	vol	NOUN
ap-990	36	35	.	.	PUNCT
ap-990	37	1	47	47	NUM
ap-990	37	2	no	no	NOUN
ap-990	37	3	.	.	PUNCT
ap-990	38	1	4–5/2007	4–5/2007	NUM
ap-990	38	2	speaker	speaker	NOUN
ap-990	38	3	non	non	ADJ
ap-990	38	4	-	-	ADJ
ap-990	38	5	speech	speech	ADJ
ap-990	38	6	event	event	NOUN
ap-990	38	7	recognition	recognition	NOUN
ap-990	38	8	with	with	ADP
ap-990	38	9	standard	standard	ADJ
ap-990	38	10	speech	speech	NOUN
ap-990	38	11	datasets	dataset	NOUN
ap-990	38	12	j.	j.	PROPN
ap-990	38	13	rajnoha	rajnoha	PROPN
ap-990	38	14	a	a	DET
ap-990	38	15	non	non	ADJ
ap-990	38	16	-	-	ADJ
ap-990	38	17	speech	speech	ADJ
ap-990	38	18	event	event	NOUN
ap-990	38	19	modelling	model	VERB
ap-990	38	20	approach	approach	NOUN
ap-990	38	21	to	to	ADP
ap-990	38	22	speech	speech	NOUN
ap-990	38	23	recognition	recognition	NOUN
ap-990	38	24	is	be	AUX
ap-990	38	25	presented	present	VERB
ap-990	38	26	in	in	ADP
ap-990	38	27	this	this	DET
ap-990	38	28	paper	paper	NOUN
ap-990	38	29	.	.	PUNCT
ap-990	39	1	a	a	DET
ap-990	39	2	speaker	speaker	NOUN
ap-990	39	3	independent	independent	ADJ
ap-990	39	4	spoken	spoken	ADJ
ap-990	39	5	czech	czech	PROPN
ap-990	39	6	digit	digit	NOUN
ap-990	39	7	recogniser	recogniser	NOUN
ap-990	39	8	is	be	AUX
ap-990	39	9	used	use	VERB
ap-990	39	10	for	for	ADP
ap-990	39	11	this	this	DET
ap-990	39	12	purpose	purpose	NOUN
ap-990	39	13	,	,	PUNCT
ap-990	39	14	and	and	CCONJ
ap-990	39	15	speaker	speaker	NOUN
ap-990	39	16	generated	generate	VERB
ap-990	39	17	non	non	ADJ
ap-990	39	18	-	-	ADJ
ap-990	39	19	speech	speech	ADJ
ap-990	39	20	events	event	NOUN
ap-990	39	21	are	be	AUX
ap-990	39	22	modelled	model	VERB
ap-990	39	23	.	.	PUNCT
ap-990	40	1	because	because	SCONJ
ap-990	40	2	it	it	PRON
ap-990	40	3	is	be	AUX
ap-990	40	4	important	important	ADJ
ap-990	40	5	for	for	SCONJ
ap-990	40	6	the	the	DET
ap-990	40	7	recogniser	recogniser	NOUN
ap-990	40	8	to	to	PART
ap-990	40	9	be	be	AUX
ap-990	40	10	trained	train	VERB
ap-990	40	11	on	on	ADP
ap-990	40	12	suitable	suitable	ADJ
ap-990	40	13	data	datum	NOUN
ap-990	40	14	,	,	PUNCT
ap-990	40	15	the	the	DET
ap-990	40	16	paper	paper	NOUN
ap-990	40	17	shows	show	VERB
ap-990	40	18	some	some	DET
ap-990	40	19	factors	factor	NOUN
ap-990	40	20	that	that	PRON
ap-990	40	21	influence	influence	VERB
ap-990	40	22	the	the	DET
ap-990	40	23	occurrence	occurrence	NOUN
ap-990	40	24	of	of	ADP
ap-990	40	25	the	the	DET
ap-990	40	26	modeled	model	VERB
ap-990	40	27	non	non	ADJ
ap-990	40	28	-	-	ADJ
ap-990	40	29	speech	speech	ADJ
ap-990	40	30	events	event	NOUN
ap-990	40	31	in	in	ADP
ap-990	40	32	the	the	DET
ap-990	40	33	training	training	NOUN
ap-990	40	34	database	database	NOUN
ap-990	40	35	.	.	PUNCT
ap-990	41	1	some	some	DET
ap-990	41	2	results	result	NOUN
ap-990	41	3	achieved	achieve	VERB
ap-990	41	4	on	on	ADP
ap-990	41	5	the	the	DET
ap-990	41	6	analysed	analyse	VERB
ap-990	41	7	training	training	NOUN
ap-990	41	8	database	database	NOUN
ap-990	41	9	are	be	AUX
ap-990	41	10	then	then	ADV
ap-990	41	11	shown	show	VERB
ap-990	41	12	.	.	PUNCT
ap-990	42	1	in	in	ADP
ap-990	42	2	the	the	DET
ap-990	42	3	experiments	experiment	NOUN
ap-990	42	4	on	on	ADP
ap-990	42	5	forced	forced	ADJ
ap-990	42	6	alignment	alignment	NOUN
ap-990	42	7	the	the	DET
ap-990	42	8	recogniser	recogniser	NOUN
ap-990	42	9	eliminates	eliminate	VERB
ap-990	42	10	almost	almost	ADV
ap-990	42	11	all	all	DET
ap-990	42	12	the	the	DET
ap-990	42	13	insertion	insertion	NOUN
ap-990	42	14	error	error	NOUN
ap-990	42	15	,	,	PUNCT
ap-990	42	16	which	which	PRON
ap-990	42	17	is	be	AUX
ap-990	42	18	a	a	DET
ap-990	42	19	promising	promising	ADJ
ap-990	42	20	property	property	NOUN
ap-990	42	21	for	for	ADP
ap-990	42	22	subsequent	subsequent	ADJ
ap-990	42	23	training	training	NOUN
ap-990	42	24	.	.	PUNCT
ap-990	43	1	however	however	ADV
ap-990	43	2	,	,	PUNCT
ap-990	43	3	experiments	experiment	NOUN
ap-990	43	4	with	with	ADP
ap-990	43	5	a	a	DET
ap-990	43	6	different	different	ADJ
ap-990	43	7	basis	basis	NOUN
ap-990	43	8	for	for	SCONJ
ap-990	43	9	the	the	DET
ap-990	43	10	non	non	ADJ
ap-990	43	11	-	-	ADJ
ap-990	43	12	speech	speech	ADJ
ap-990	43	13	event	event	NOUN
ap-990	43	14	models	model	NOUN
ap-990	43	15	provide	provide	VERB
ap-990	43	16	almost	almost	ADV
ap-990	43	17	the	the	DET
ap-990	43	18	same	same	ADJ
ap-990	43	19	results	result	NOUN
ap-990	43	20	,	,	PUNCT
ap-990	43	21	so	so	CCONJ
ap-990	43	22	the	the	DET
ap-990	43	23	difference	difference	NOUN
ap-990	43	24	seems	seem	VERB
ap-990	43	25	to	to	PART
ap-990	43	26	be	be	AUX
ap-990	43	27	not	not	PART
ap-990	43	28	so	so	ADV
ap-990	43	29	significant	significant	ADJ
ap-990	43	30	for	for	ADP
ap-990	43	31	recognition	recognition	NOUN
ap-990	43	32	.	.	PUNCT
ap-990	44	1	keywords	keyword	NOUN
ap-990	44	2	:	:	PUNCT
ap-990	44	3	speech	speech	NOUN
ap-990	44	4	recognition	recognition	NOUN
ap-990	44	5	,	,	PUNCT
ap-990	44	6	digit	digit	NOUN
ap-990	44	7	recognition	recognition	NOUN
ap-990	44	8	,	,	PUNCT
ap-990	44	9	non	non	ADJ
ap-990	44	10	-	-	ADJ
ap-990	44	11	speech	speech	ADJ
ap-990	44	12	events	event	NOUN
ap-990	44	13	,	,	PUNCT
ap-990	44	14	training	training	NOUN
ap-990	44	15	database	database	NOUN
ap-990	44	16	,	,	PUNCT
ap-990	44	17	forced	force	VERB
ap-990	44	18	alignment	alignment	NOUN
ap-990	44	19	.	.	PUNCT
ap-990	45	1	scription	scription	NOUN
ap-990	45	2	.	.	PUNCT
ap-990	46	1	the	the	DET
ap-990	46	2	analysis	analysis	NOUN
ap-990	46	3	in	in	ADP
ap-990	46	4	2.2	2.2	NUM
ap-990	46	5	shows	show	NOUN
ap-990	46	6	that	that	SCONJ
ap-990	46	7	higher	high	ADJ
ap-990	46	8	snr	snr	NOUN
ap-990	46	9	(	(	PUNCT
ap-990	46	10	lower	low	ADJ
ap-990	46	11	noise	noise	NOUN
ap-990	46	12	level	level	NOUN
ap-990	46	13	)	)	PUNCT
ap-990	46	14	in	in	ADP
ap-990	46	15	speech	speech	NOUN
ap-990	46	16	leads	lead	VERB
ap-990	46	17	to	to	ADP
ap-990	46	18	lower	low	ADJ
ap-990	46	19	occurrence	occurrence	NOUN
ap-990	46	20	of	of	ADP
ap-990	46	21	the	the	DET
ap-990	46	22	events	event	NOUN
ap-990	46	23	.	.	PUNCT
ap-990	47	1	some	some	DET
ap-990	47	2	compromise	compromise	NOUN
ap-990	47	3	is	be	AUX
ap-990	47	4	therefore	therefore	ADV
ap-990	47	5	needed	need	VERB
ap-990	47	6	in	in	ADP
ap-990	47	7	the	the	DET
ap-990	47	8	choice	choice	NOUN
ap-990	47	9	between	between	ADP
ap-990	47	10	clean	clean	ADJ
ap-990	47	11	speech	speech	NOUN
ap-990	47	12	(	(	PUNCT
ap-990	47	13	more	more	ADV
ap-990	47	14	understandable	understandable	ADJ
ap-990	47	15	for	for	ADP
ap-990	47	16	the	the	DET
ap-990	47	17	recogniser	recogniser	NOUN
ap-990	47	18	)	)	PUNCT
ap-990	47	19	and	and	CCONJ
ap-990	47	20	the	the	DET
ap-990	47	21	number	number	NOUN
ap-990	47	22	of	of	ADP
ap-990	47	23	non	non	ADJ
ap-990	47	24	-	-	ADJ
ap-990	47	25	speech	speech	ADJ
ap-990	47	26	events	event	NOUN
ap-990	47	27	for	for	ADP
ap-990	47	28	training	train	VERB
ap-990	47	29	the	the	DET
ap-990	47	30	non	non	ADJ
ap-990	47	31	-	-	ADJ
ap-990	47	32	speech	speech	ADJ
ap-990	47	33	event	event	NOUN
ap-990	47	34	robust	robust	ADJ
ap-990	47	35	recogniser	recogniser	NOUN
ap-990	47	36	.	.	PUNCT
ap-990	48	1	the	the	DET
ap-990	48	2	reason	reason	NOUN
ap-990	48	3	for	for	ADP
ap-990	48	4	the	the	DET
ap-990	48	5	low	low	ADJ
ap-990	48	6	number	number	NOUN
ap-990	48	7	of	of	ADP
ap-990	48	8	non	non	ADJ
ap-990	48	9	-	-	ADJ
ap-990	48	10	speech	speech	ADJ
ap-990	48	11	events	event	NOUN
ap-990	48	12	in	in	ADP
ap-990	48	13	the	the	DET
ap-990	48	14	tem	tem	NOUN
ap-990	48	15	-	-	PUNCT
ap-990	48	16	training	training	NOUN
ap-990	48	17	subset	subset	NOUN
ap-990	48	18	is	be	AUX
ap-990	48	19	that	that	SCONJ
ap-990	48	20	the	the	DET
ap-990	48	21	training	training	NOUN
ap-990	48	22	subset	subset	NOUN
ap-990	48	23	is	be	AUX
ap-990	48	24	only	only	ADV
ap-990	48	25	a	a	DET
ap-990	48	26	small	small	ADJ
ap-990	48	27	fragment	fragment	NOUN
ap-990	48	28	of	of	ADP
ap-990	48	29	the	the	DET
ap-990	48	30	whole	whole	ADJ
ap-990	48	31	tem	tem	NOUN
ap-990	48	32	dataset	dataset	VERB
ap-990	48	33	.	.	PUNCT
ap-990	49	1	only	only	ADV
ap-990	49	2	the	the	DET
ap-990	49	3	standing	stand	VERB
ap-990	49	4	car	car	NOUN
ap-990	49	5	items	item	NOUN
ap-990	49	6	were	be	AUX
ap-990	49	7	taken	take	VERB
ap-990	49	8	for	for	ADP
ap-990	49	9	training	training	NOUN
ap-990	49	10	purposes	purpose	NOUN
ap-990	49	11	.	.	PUNCT
ap-990	50	1	the	the	DET
ap-990	50	2	datasets	dataset	NOUN
ap-990	50	3	also	also	ADV
ap-990	50	4	include	include	VERB
ap-990	50	5	some	some	DET
ap-990	50	6	records	record	NOUN
ap-990	50	7	that	that	PRON
ap-990	50	8	are	be	AUX
ap-990	50	9	not	not	PART
ap-990	50	10	suitable	suitable	ADJ
ap-990	50	11	for	for	ADP
ap-990	50	12	standard	standard	ADJ
ap-990	50	13	phoneme	phoneme	NOUN
ap-990	50	14	training	training	NOUN
ap-990	50	15	,	,	PUNCT
ap-990	50	16	e.g.	e.g.	ADV
ap-990	50	17	web	web	NOUN
ap-990	50	18	-	-	ADJ
ap-990	50	19	page	page	NOUN
ap-990	50	20	addresses	address	NOUN
ap-990	50	21	or	or	CCONJ
ap-990	50	22	spelled	spell	VERB
ap-990	50	23	utterances	utterance	NOUN
ap-990	50	24	.	.	PUNCT
ap-990	51	1	while	while	SCONJ
ap-990	51	2	the	the	DET
ap-990	51	3	average	average	ADJ
ap-990	51	4	occurrence	occurrence	NOUN
ap-990	51	5	of	of	ADP
ap-990	51	6	a	a	DET
ap-990	51	7	non	non	ADJ
ap-990	51	8	-	-	ADJ
ap-990	51	9	speech	speech	ADJ
ap-990	51	10	event	event	NOUN
ap-990	51	11	is	be	AUX
ap-990	51	12	0.64	0.64	NUM
ap-990	51	13	events	event	NOUN
ap-990	51	14	per	per	ADP
ap-990	51	15	one	one	NUM
ap-990	51	16	utterance	utterance	NOUN
ap-990	51	17	in	in	ADP
ap-990	51	18	the	the	DET
ap-990	51	19	whole	whole	ADJ
ap-990	51	20	spee	spee	NOUN
ap-990	51	21	dataset	dataset	NOUN
ap-990	51	22	and	and	CCONJ
ap-990	51	23	0.56	0.56	NUM
ap-990	51	24	events	event	NOUN
ap-990	51	25	per	per	ADP
ap-990	51	26	utterance	utterance	NOUN
ap-990	51	27	in	in	ADP
ap-990	51	28	the	the	DET
ap-990	51	29	training	training	NOUN
ap-990	51	30	subset	subset	NOUN
ap-990	51	31	,	,	PUNCT
ap-990	51	32	the	the	DET
ap-990	51	33	average	average	ADJ
ap-990	51	34	rate	rate	NOUN
ap-990	51	35	for	for	ADP
ap-990	51	36	web	web	NOUN
ap-990	51	37	-	-	NOUN
ap-990	51	38	page	page	NOUN
ap-990	51	39	utterances	utterance	NOUN
ap-990	51	40	is	be	AUX
ap-990	51	41	0.89	0.89	NUM
ap-990	51	42	.	.	PUNCT
ap-990	52	1	however	however	ADV
ap-990	52	2	,	,	PUNCT
ap-990	52	3	these	these	DET
ap-990	52	4	records	record	NOUN
ap-990	52	5	were	be	AUX
ap-990	52	6	not	not	PART
ap-990	52	7	used	use	VERB
ap-990	52	8	in	in	ADP
ap-990	52	9	the	the	DET
ap-990	52	10	training	training	NOUN
ap-990	52	11	phase	phase	NOUN
ap-990	52	12	,	,	PUNCT
ap-990	52	13	which	which	PRON
ap-990	52	14	decreases	decrease	VERB
ap-990	52	15	the	the	DET
ap-990	52	16	number	number	NOUN
ap-990	52	17	of	of	ADP
ap-990	52	18	non	non	ADJ
ap-990	52	19	-	-	ADJ
ap-990	52	20	speech	speech	ADJ
ap-990	52	21	events	event	NOUN
ap-990	52	22	in	in	ADP
ap-990	52	23	the	the	DET
ap-990	52	24	training	training	NOUN
ap-990	52	25	subset	subset	NOUN
ap-990	52	26	.	.	PUNCT
ap-990	53	1	the	the	DET
ap-990	53	2	analyses	analysis	NOUN
ap-990	53	3	follow	follow	VERB
ap-990	53	4	.	.	PUNCT
ap-990	54	1	these	these	PRON
ap-990	54	2	try	try	VERB
ap-990	54	3	to	to	PART
ap-990	54	4	find	find	VERB
ap-990	54	5	groups	group	NOUN
ap-990	54	6	of	of	ADP
ap-990	54	7	records	record	NOUN
ap-990	54	8	with	with	ADP
ap-990	54	9	some	some	DET
ap-990	54	10	property	property	NOUN
ap-990	54	11	that	that	PRON
ap-990	54	12	is	be	AUX
ap-990	54	13	important	important	ADJ
ap-990	54	14	in	in	ADP
ap-990	54	15	the	the	DET
ap-990	54	16	speaker	speaker	NOUN
ap-990	54	17	non	non	ADJ
ap-990	54	18	-	-	ADJ
ap-990	54	19	speech	speech	ADJ
ap-990	54	20	event	event	NOUN
ap-990	54	21	distribution	distribution	NOUN
ap-990	54	22	in	in	ADP
ap-990	54	23	the	the	DET
ap-990	54	24	training	training	NOUN
ap-990	54	25	subset	subset	NOUN
ap-990	54	26	.	.	PUNCT
ap-990	55	1	this	this	PRON
ap-990	55	2	can	can	AUX
ap-990	55	3	help	help	VERB
ap-990	55	4	to	to	PART
ap-990	55	5	discover	discover	VERB
ap-990	55	6	some	some	DET
ap-990	55	7	inefficiency	inefficiency	NOUN
ap-990	55	8	in	in	ADP
ap-990	55	9	the	the	DET
ap-990	55	10	non	non	ADJ
ap-990	55	11	-	-	ADJ
ap-990	55	12	speech	speech	ADJ
ap-990	55	13	event	event	NOUN
ap-990	55	14	training	training	NOUN
ap-990	55	15	process	process	NOUN
ap-990	55	16	.	.	PUNCT
ap-990	56	1	2.1	2.1	NUM
ap-990	56	2	event	event	NOUN
ap-990	56	3	distribution	distribution	NOUN
ap-990	56	4	in	in	ADP
ap-990	56	5	speaker	speaker	NOUN
ap-990	56	6	’s	’s	PART
ap-990	56	7	age	age	NOUN
ap-990	56	8	the	the	DET
ap-990	56	9	datasets	dataset	NOUN
ap-990	56	10	include	include	VERB
ap-990	56	11	some	some	DET
ap-990	56	12	basic	basic	ADJ
ap-990	56	13	information	information	NOUN
ap-990	56	14	about	about	ADP
ap-990	56	15	the	the	DET
ap-990	56	16	speaker	speaker	NOUN
ap-990	56	17	.	.	PUNCT
ap-990	57	1	this	this	PRON
ap-990	57	2	can	can	AUX
ap-990	57	3	help	help	VERB
ap-990	57	4	us	we	PRON
ap-990	57	5	to	to	PART
ap-990	57	6	find	find	VERB
ap-990	57	7	whether	whether	SCONJ
ap-990	57	8	some	some	DET
ap-990	57	9	group	group	NOUN
ap-990	57	10	of	of	ADP
ap-990	57	11	people	people	NOUN
ap-990	57	12	can	can	AUX
ap-990	57	13	influence	influence	VERB
ap-990	57	14	the	the	DET
ap-990	57	15	training	training	NOUN
ap-990	57	16	dataset	dataset	NOUN
ap-990	57	17	in	in	ADP
ap-990	57	18	some	some	DET
ap-990	57	19	way	way	NOUN
ap-990	57	20	.	.	PUNCT
ap-990	58	1	one	one	NUM
ap-990	58	2	such	such	ADJ
ap-990	58	3	kind	kind	NOUN
ap-990	58	4	of	of	ADP
ap-990	58	5	information	information	NOUN
ap-990	58	6	is	be	AUX
ap-990	58	7	the	the	DET
ap-990	58	8	age	age	NOUN
ap-990	58	9	-	-	PUNCT
ap-990	58	10	class	class	NOUN
ap-990	58	11	distribution	distribution	NOUN
ap-990	58	12	and	and	CCONJ
ap-990	58	13	the	the	DET
ap-990	58	14	number	number	NOUN
ap-990	58	15	of	of	ADP
ap-990	58	16	speaker	speaker	NOUN
ap-990	58	17	non	non	ADJ
ap-990	58	18	-	-	NOUN
ap-990	58	19	speech	speech	ADJ
ap-990	58	20	events	event	NOUN
ap-990	58	21	for	for	ADP
ap-990	58	22	different	different	ADJ
ap-990	58	23	age	age	NOUN
ap-990	58	24	-	-	PUNCT
ap-990	58	25	classes	class	NOUN
ap-990	58	26	.	.	PUNCT
ap-990	59	1	figure	figure	NOUN
ap-990	59	2	2	2	NUM
ap-990	59	3	shows	show	VERB
ap-990	59	4	that	that	SCONJ
ap-990	59	5	the	the	DET
ap-990	59	6	presence	presence	NOUN
ap-990	59	7	of	of	ADP
ap-990	59	8	speaker	speaker	NOUN
ap-990	59	9	non	non	ADJ
ap-990	59	10	-	-	ADJ
ap-990	59	11	speech	speech	ADJ
ap-990	59	12	events	event	NOUN
ap-990	59	13	is	be	AUX
ap-990	59	14	not	not	PART
ap-990	59	15	much	much	ADV
ap-990	59	16	influenced	influence	VERB
ap-990	59	17	by	by	ADP
ap-990	59	18	the	the	DET
ap-990	59	19	speakers	speaker	NOUN
ap-990	59	20	age	age	NOUN
ap-990	59	21	till	till	SCONJ
ap-990	59	22	about	about	ADP
ap-990	59	23	65	65	NUM
ap-990	59	24	years	year	NOUN
ap-990	59	25	,	,	PUNCT
ap-990	59	26	and	and	CCONJ
ap-990	59	27	only	only	ADV
ap-990	59	28	for	for	ADP
ap-990	59	29	higher	high	ADJ
ap-990	59	30	age	age	NOUN
ap-990	59	31	the	the	DET
ap-990	59	32	amount	amount	NOUN
ap-990	59	33	rises	rise	VERB
ap-990	59	34	.	.	PUNCT
ap-990	60	1	there	there	PRON
ap-990	60	2	are	be	VERB
ap-990	60	3	not	not	PART
ap-990	60	4	many	many	ADJ
ap-990	60	5	speakers	speaker	NOUN
ap-990	60	6	from	from	ADP
ap-990	60	7	the	the	DET
ap-990	60	8	65	65	NUM
ap-990	60	9	+	+	CCONJ
ap-990	60	10	age	age	NOUN
ap-990	60	11	-	-	PUNCT
ap-990	60	12	class	class	NOUN
ap-990	60	13	in	in	ADP
ap-990	60	14	the	the	DET
ap-990	60	15	datasets	dataset	NOUN
ap-990	60	16	(	(	PUNCT
ap-990	60	17	fig	fig	NOUN
ap-990	60	18	.	.	NOUN
ap-990	60	19	1	1	NUM
ap-990	60	20	)	)	PUNCT
ap-990	60	21	,	,	PUNCT
ap-990	60	22	and	and	CCONJ
ap-990	60	23	this	this	DET
ap-990	60	24	increase	increase	NOUN
ap-990	60	25	is	be	AUX
ap-990	60	26	not	not	PART
ap-990	60	27	high	high	ADJ
ap-990	60	28	enough	enough	ADV
ap-990	60	29	to	to	PART
ap-990	60	30	cause	cause	VERB
ap-990	60	31	harmful	harmful	ADJ
ap-990	60	32	effects	effect	NOUN
ap-990	60	33	,	,	PUNCT
ap-990	60	34	like	like	ADP
ap-990	60	35	training	train	VERB
ap-990	60	36	the	the	DET
ap-990	60	37	recogniser	recogniser	NOUN
ap-990	60	38	for	for	ADP
ap-990	60	39	a	a	DET
ap-990	60	40	specific	specific	ADJ
ap-990	60	41	kind	kind	NOUN
ap-990	60	42	of	of	ADP
ap-990	60	43	event	event	NOUN
ap-990	60	44	only	only	ADV
ap-990	60	45	.	.	PUNCT
ap-990	61	1	2.2	2.2	NUM
ap-990	61	2	event	event	NOUN
ap-990	61	3	distribution	distribution	NOUN
ap-990	61	4	in	in	ADP
ap-990	61	5	different	different	ADJ
ap-990	61	6	noise	noise	NOUN
ap-990	61	7	levels	level	NOUN
ap-990	61	8	the	the	DET
ap-990	61	9	spee	spee	PROPN
ap-990	61	10	database	database	NOUN
ap-990	61	11	includes	include	VERB
ap-990	61	12	an	an	DET
ap-990	61	13	snr	snr	NOUN
ap-990	61	14	estimation	estimation	NOUN
ap-990	61	15	for	for	ADP
ap-990	61	16	the	the	DET
ap-990	61	17	records	record	NOUN
ap-990	61	18	.	.	PUNCT
ap-990	62	1	it	it	PRON
ap-990	62	2	enables	enable	VERB
ap-990	62	3	the	the	DET
ap-990	62	4	user	user	NOUN
ap-990	62	5	to	to	PART
ap-990	62	6	analyse	analyse	VERB
ap-990	62	7	the	the	DET
ap-990	62	8	influence	influence	NOUN
ap-990	62	9	of	of	ADP
ap-990	62	10	a	a	DET
ap-990	62	11	noisy	noisy	ADJ
ap-990	62	12	environment	environment	NOUN
ap-990	62	13	on	on	ADP
ap-990	62	14	the	the	DET
ap-990	62	15	occurrence	occurrence	NOUN
ap-990	62	16	of	of	ADP
ap-990	62	17	a	a	DET
ap-990	62	18	non	non	ADJ
ap-990	62	19	-	-	ADJ
ap-990	62	20	speech	speech	ADJ
ap-990	62	21	event	event	NOUN
ap-990	62	22	in	in	ADP
ap-990	62	23	speech	speech	NOUN
ap-990	62	24	.	.	PUNCT
ap-990	63	1	there	there	PRON
ap-990	63	2	is	be	VERB
ap-990	63	3	no	no	DET
ap-990	63	4	snr	snr	NOUN
ap-990	63	5	estimate	estimate	NOUN
ap-990	63	6	for	for	ADP
ap-990	63	7	the	the	DET
ap-990	63	8	tem	tem	PROPN
ap-990	63	9	database	database	NOUN
ap-990	63	10	,	,	PUNCT
ap-990	63	11	but	but	CCONJ
ap-990	63	12	the	the	DET
ap-990	63	13	information	information	NOUN
ap-990	63	14	on	on	ADP
ap-990	63	15	car	car	NOUN
ap-990	63	16	type	type	NOUN
ap-990	63	17	and	and	CCONJ
ap-990	63	18	engine	engine	NOUN
ap-990	63	19	state	state	NOUN
ap-990	63	20	can	can	AUX
ap-990	63	21	also	also	ADV
ap-990	63	22	partially	partially	ADV
ap-990	63	23	describe	describe	VERB
ap-990	63	24	the	the	DET
ap-990	63	25	environment	environment	NOUN
ap-990	63	26	.	.	PUNCT
ap-990	64	1	the	the	DET
ap-990	64	2	graph	graph	NOUN
ap-990	64	3	in	in	ADP
ap-990	64	4	fig	fig	NOUN
ap-990	64	5	.	.	PUNCT
ap-990	65	1	3	3	NUM
ap-990	65	2	shows	show	VERB
ap-990	65	3	that	that	SCONJ
ap-990	65	4	the	the	DET
ap-990	65	5	distribution	distribution	NOUN
ap-990	65	6	of	of	ADP
ap-990	65	7	the	the	DET
ap-990	65	8	average	average	ADJ
ap-990	65	9	non	non	ADJ
ap-990	65	10	-	-	ADJ
ap-990	65	11	speech	speech	ADJ
ap-990	65	12	event	event	NOUN
ap-990	65	13	rate	rate	NOUN
ap-990	65	14	has	have	VERB
ap-990	65	15	its	its	PRON
ap-990	65	16	maximum	maximum	NOUN
ap-990	65	17	in	in	ADP
ap-990	65	18	the	the	DET
ap-990	65	19	snr	snr	PROPN
ap-990	65	20	group	group	NOUN
ap-990	65	21	between	between	ADP
ap-990	65	22	10	10	NUM
ap-990	65	23	and	and	CCONJ
ap-990	65	24	15db	15db	NOUN
ap-990	65	25	.	.	PUNCT
ap-990	66	1	for	for	ADP
ap-990	66	2	lower	low	ADJ
ap-990	66	3	snr	snr	NOUN
ap-990	66	4	,	,	PUNCT
ap-990	66	5	the	the	DET
ap-990	66	6	rate	rate	NOUN
ap-990	66	7	falls	fall	VERB
ap-990	66	8	mainly	mainly	ADV
ap-990	66	9	because	because	SCONJ
ap-990	66	10	of	of	ADP
ap-990	66	11	the	the	DET
ap-990	66	12	noise	noise	NOUN
ap-990	66	13	that	that	PRON
ap-990	66	14	covers	cover	VERB
ap-990	66	15	the	the	DET
ap-990	66	16	event	event	NOUN
ap-990	66	17	.	.	PUNCT
ap-990	67	1	a	a	DET
ap-990	67	2	similar	similar	ADJ
ap-990	67	3	effect	effect	NOUN
ap-990	67	4	can	can	AUX
ap-990	67	5	be	be	AUX
ap-990	67	6	seen	see	VERB
ap-990	67	7	in	in	ADP
ap-990	67	8	tem	tem	PROPN
ap-990	67	9	,	,	PUNCT
ap-990	67	10	where	where	SCONJ
ap-990	67	11	the	the	DET
ap-990	67	12	items	item	NOUN
ap-990	67	13	without	without	ADP
ap-990	67	14	a	a	DET
ap-990	67	15	running	running	NOUN
ap-990	67	16	engine	engine	NOUN
ap-990	67	17	in	in	ADP
ap-990	67	18	the	the	DET
ap-990	67	19	background	background	NOUN
ap-990	67	20	contain	contain	VERB
ap-990	67	21	more	more	ADJ
ap-990	67	22	non	non	ADJ
ap-990	67	23	-	-	ADJ
ap-990	67	24	speech	speech	ADJ
ap-990	67	25	event	event	NOUN
ap-990	67	26	marks	mark	NOUN
ap-990	67	27	.	.	PUNCT
ap-990	68	1	for	for	ADP
ap-990	68	2	higher	high	ADJ
ap-990	68	3	snr	snr	NOUN
ap-990	68	4	ranges	range	VERB
ap-990	68	5	the	the	DET
ap-990	68	6	rate	rate	NOUN
ap-990	68	7	also	also	ADV
ap-990	68	8	falls	fall	VERB
ap-990	68	9	.	.	PUNCT
ap-990	69	1	in	in	ADP
ap-990	69	2	this	this	DET
ap-990	69	3	environment	environment	NOUN
ap-990	69	4	the	the	DET
ap-990	69	5	108	108	NUM
ap-990	69	6	©	©	PROPN
ap-990	69	7	czech	czech	PROPN
ap-990	69	8	technical	technical	PROPN
ap-990	69	9	university	university	PROPN
ap-990	69	10	publishing	publishing	NOUN
ap-990	69	11	house	house	NOUN
ap-990	69	12	http://ctn.cvut.cz/ap/	http://ctn.cvut.cz/ap/	PROPN
ap-990	69	13	acta	acta	PROPN
ap-990	69	14	polytechnica	polytechnica	PROPN
ap-990	69	15	vol	vol	NOUN
ap-990	69	16	.	.	PUNCT
ap-990	70	1	47	47	NUM
ap-990	70	2	no	no	NOUN
ap-990	70	3	.	.	PUNCT
ap-990	71	1	4–5/2007	4–5/2007	NUM
ap-990	71	2	utterances	utterance	NOUN
ap-990	71	3	spk	spk	NOUN
ap-990	71	4	-	-	PUNCT
ap-990	71	5	event	event	NOUN
ap-990	71	6	fil	fil	NOUN
ap-990	71	7	-	-	NOUN
ap-990	71	8	event	event	NOUN
ap-990	71	9	spee	spee	NOUN
ap-990	71	10	-	-	PUNCT
ap-990	71	11	all	all	PRON
ap-990	71	12	180213	180213	NUM
ap-990	71	13	108474	108474	NUM
ap-990	71	14	7382	7382	NUM
ap-990	71	15	spee	spee	NOUN
ap-990	71	16	-	-	PUNCT
ap-990	71	17	train	train	NOUN
ap-990	71	18	63024	63024	NUM
ap-990	71	19	33138	33138	NUM
ap-990	71	20	1856	1856	NUM
ap-990	71	21	tem	tem	PROPN
ap-990	71	22	-	-	PUNCT
ap-990	71	23	all	all	PRON
ap-990	71	24	221318	221318	NUM
ap-990	71	25	46742	46742	NUM
ap-990	71	26	691	691	NUM
ap-990	71	27	tem	tem	PROPN
ap-990	71	28	-	-	PUNCT
ap-990	71	29	train	train	NOUN
ap-990	71	30	38391	38391	NUM
ap-990	71	31	11532	11532	NUM
ap-990	71	32	153	153	NUM
ap-990	71	33	table	table	NOUN
ap-990	71	34	1	1	NUM
ap-990	71	35	:	:	PUNCT
ap-990	71	36	number	number	NOUN
ap-990	71	37	of	of	ADP
ap-990	71	38	non	non	ADJ
ap-990	71	39	-	-	ADJ
ap-990	71	40	speech	speech	ADJ
ap-990	71	41	events	event	NOUN
ap-990	71	42	fig	fig	NOUN
ap-990	71	43	.	.	PUNCT
ap-990	72	1	1	1	NUM
ap-990	72	2	:	:	PUNCT
ap-990	72	3	age	age	NOUN
ap-990	72	4	-	-	PUNCT
ap-990	72	5	class	class	NOUN
ap-990	72	6	distribution	distribution	NOUN
ap-990	72	7	in	in	ADP
ap-990	72	8	the	the	DET
ap-990	72	9	datasets	dataset	NOUN
ap-990	72	10	fig	fig	NOUN
ap-990	72	11	.	.	PUNCT
ap-990	73	1	2	2	NUM
ap-990	73	2	:	:	PUNCT
ap-990	73	3	non	non	ADJ
ap-990	73	4	-	-	ADJ
ap-990	73	5	speech	speech	ADJ
ap-990	73	6	events	event	NOUN
ap-990	73	7	distribution	distribution	NOUN
ap-990	73	8	in	in	ADP
ap-990	73	9	age	age	NOUN
ap-990	73	10	for	for	ADP
ap-990	73	11	the	the	DET
ap-990	73	12	speech	speech	NOUN
ap-990	73	13	datasets	dataset	NOUN
ap-990	73	14	fig	fig	NOUN
ap-990	73	15	.	.	PUNCT
ap-990	74	1	3	3	NUM
ap-990	74	2	:	:	PUNCT
ap-990	74	3	non	non	ADJ
ap-990	74	4	-	-	ADJ
ap-990	74	5	speech	speech	ADJ
ap-990	74	6	events	event	NOUN
ap-990	74	7	in	in	ADP
ap-990	74	8	the	the	DET
ap-990	74	9	training	training	NOUN
ap-990	74	10	spee	spee	NOUN
ap-990	74	11	dataset	dataset	VERB
ap-990	74	12	in	in	ADP
ap-990	74	13	different	different	ADJ
ap-990	74	14	noisy	noisy	ADJ
ap-990	74	15	environment	environment	NOUN
ap-990	74	16	speaker	speaker	NOUN
ap-990	74	17	is	be	AUX
ap-990	74	18	not	not	PART
ap-990	74	19	disturbed	disturb	VERB
ap-990	74	20	by	by	ADP
ap-990	74	21	the	the	DET
ap-990	74	22	environmental	environmental	ADJ
ap-990	74	23	noise	noise	NOUN
ap-990	74	24	so	so	SCONJ
ap-990	74	25	he	he	PRON
ap-990	74	26	pronounces	pronounce	VERB
ap-990	74	27	properly	properly	ADV
ap-990	74	28	and	and	CCONJ
ap-990	74	29	this	this	PRON
ap-990	74	30	leads	lead	VERB
ap-990	74	31	to	to	ADP
ap-990	74	32	a	a	DET
ap-990	74	33	lower	low	ADJ
ap-990	74	34	non	non	ADJ
ap-990	74	35	-	-	ADJ
ap-990	74	36	speech	speech	ADJ
ap-990	74	37	event	event	NOUN
ap-990	74	38	rate	rate	NOUN
ap-990	74	39	.	.	PUNCT
ap-990	75	1	2.3	2.3	NUM
ap-990	75	2	event	event	NOUN
ap-990	75	3	distribution	distribution	NOUN
ap-990	75	4	for	for	ADP
ap-990	75	5	different	different	ADJ
ap-990	75	6	workgroups	workgroup	NOUN
ap-990	75	7	the	the	DET
ap-990	75	8	realization	realization	NOUN
ap-990	75	9	team	team	NOUN
ap-990	75	10	for	for	ADP
ap-990	75	11	creating	create	VERB
ap-990	75	12	the	the	DET
ap-990	75	13	two	two	NUM
ap-990	75	14	databases	database	NOUN
ap-990	75	15	was	be	AUX
ap-990	75	16	divided	divide	VERB
ap-990	75	17	into	into	ADP
ap-990	75	18	two	two	NUM
ap-990	75	19	parts	part	NOUN
ap-990	75	20	with	with	ADP
ap-990	75	21	different	different	ADJ
ap-990	75	22	supervisors	supervisor	NOUN
ap-990	75	23	and	and	CCONJ
ap-990	75	24	workplaces	workplace	NOUN
ap-990	75	25	within	within	ADP
ap-990	75	26	the	the	DET
ap-990	75	27	czech	czech	PROPN
ap-990	75	28	republic	republic	NOUN
ap-990	75	29	.	.	PUNCT
ap-990	76	1	this	this	PRON
ap-990	76	2	led	lead	VERB
ap-990	76	3	to	to	ADP
ap-990	76	4	a	a	DET
ap-990	76	5	different	different	ADJ
ap-990	76	6	dialect	dialect	NOUN
ap-990	76	7	distribution	distribution	NOUN
ap-990	76	8	within	within	ADP
ap-990	76	9	the	the	DET
ap-990	76	10	groups	group	NOUN
ap-990	76	11	,	,	PUNCT
ap-990	76	12	as	as	SCONJ
ap-990	76	13	the	the	DET
ap-990	76	14	records	record	NOUN
ap-990	76	15	from	from	ADP
ap-990	76	16	the	the	DET
ap-990	76	17	second	second	ADJ
ap-990	76	18	group	group	NOUN
ap-990	76	19	were	be	AUX
ap-990	76	20	spoken	speak	VERB
ap-990	76	21	mainly	mainly	ADV
ap-990	76	22	by	by	ADP
ap-990	76	23	the	the	DET
ap-990	76	24	speakers	speaker	NOUN
ap-990	76	25	with	with	ADP
ap-990	76	26	a	a	DET
ap-990	76	27	moravian	moravian	ADJ
ap-990	76	28	dialect	dialect	NOUN
ap-990	76	29	.	.	PUNCT
ap-990	77	1	as	as	SCONJ
ap-990	77	2	shown	show	VERB
ap-990	77	3	in	in	ADP
ap-990	77	4	graph	graph	NOUN
ap-990	77	5	5	5	NUM
ap-990	77	6	,	,	PUNCT
ap-990	77	7	this	this	DET
ap-990	77	8	separation	separation	NOUN
ap-990	77	9	led	lead	VERB
ap-990	77	10	to	to	ADP
ap-990	77	11	a	a	DET
ap-990	77	12	difference	difference	NOUN
ap-990	77	13	in	in	ADP
ap-990	77	14	the	the	DET
ap-990	77	15	annotations	annotation	NOUN
ap-990	77	16	.	.	PUNCT
ap-990	78	1	unlike	unlike	ADP
ap-990	78	2	the	the	DET
ap-990	78	3	second	second	ADJ
ap-990	78	4	group	group	NOUN
ap-990	78	5	,	,	PUNCT
ap-990	78	6	the	the	DET
ap-990	78	7	workers	worker	NOUN
ap-990	78	8	in	in	ADP
ap-990	78	9	the	the	DET
ap-990	78	10	first	first	ADJ
ap-990	78	11	group	group	NOUN
ap-990	78	12	had	have	VERB
ap-990	78	13	to	to	PART
ap-990	78	14	use	use	VERB
ap-990	78	15	fewer	few	ADJ
ap-990	78	16	non	non	ADJ
ap-990	78	17	-	-	ADJ
ap-990	78	18	speech	speech	ADJ
ap-990	78	19	event	event	NOUN
ap-990	78	20	marks	mark	NOUN
ap-990	78	21	in	in	ADP
ap-990	78	22	the	the	DET
ap-990	78	23	transcription	transcription	NOUN
ap-990	78	24	,	,	PUNCT
ap-990	78	25	and	and	CCONJ
ap-990	78	26	these	these	DET
ap-990	78	27	items	item	NOUN
ap-990	78	28	have	have	VERB
ap-990	78	29	a	a	DET
ap-990	78	30	notably	notably	ADV
ap-990	78	31	lower	low	ADJ
ap-990	78	32	rate	rate	NOUN
ap-990	78	33	of	of	ADP
ap-990	78	34	non	non	ADJ
ap-990	78	35	-	-	ADJ
ap-990	78	36	speech	speech	ADJ
ap-990	78	37	event	event	NOUN
ap-990	78	38	occurrence	occurrence	NOUN
ap-990	78	39	.	.	PUNCT
ap-990	79	1	this	this	PRON
ap-990	79	2	may	may	AUX
ap-990	79	3	have	have	AUX
ap-990	79	4	caused	cause	VERB
ap-990	79	5	some	some	PRON
ap-990	79	6	of	of	ADP
ap-990	79	7	the	the	DET
ap-990	79	8	marked	mark	VERB
ap-990	79	9	non	non	ADJ
ap-990	79	10	-	-	ADJ
ap-990	79	11	speech	speech	ADJ
ap-990	79	12	events	event	NOUN
ap-990	79	13	from	from	ADP
ap-990	79	14	the	the	DET
ap-990	79	15	second	second	ADJ
ap-990	79	16	group	group	NOUN
ap-990	79	17	to	to	PART
ap-990	79	18	be	be	AUX
ap-990	79	19	less	less	ADV
ap-990	79	20	loud	loud	ADJ
ap-990	79	21	or	or	CCONJ
ap-990	79	22	less	less	ADV
ap-990	79	23	prominent	prominent	ADJ
ap-990	79	24	.	.	PUNCT
ap-990	80	1	because	because	SCONJ
ap-990	80	2	such	such	ADJ
ap-990	80	3	events	event	NOUN
ap-990	80	4	do	do	AUX
ap-990	80	5	not	not	PART
ap-990	80	6	decrease	decrease	VERB
ap-990	80	7	the	the	DET
ap-990	80	8	recognition	recognition	NOUN
ap-990	80	9	score	score	NOUN
ap-990	80	10	,	,	PUNCT
ap-990	80	11	it	it	PRON
ap-990	80	12	is	be	AUX
ap-990	80	13	quite	quite	ADV
ap-990	80	14	undesirable	undesirable	ADJ
ap-990	80	15	to	to	PART
ap-990	80	16	train	train	VERB
ap-990	80	17	the	the	DET
ap-990	80	18	recogniser	recogniser	NOUN
ap-990	80	19	to	to	PART
ap-990	80	20	separate	separate	VERB
ap-990	80	21	them	they	PRON
ap-990	80	22	from	from	ADP
ap-990	80	23	common	common	ADJ
ap-990	80	24	background	background	NOUN
ap-990	80	25	noise	noise	NOUN
ap-990	80	26	.	.	PUNCT
ap-990	81	1	if	if	SCONJ
ap-990	81	2	these	these	DET
ap-990	81	3	unimportant	unimportant	ADJ
ap-990	81	4	event	event	NOUN
ap-990	81	5	marks	mark	NOUN
ap-990	81	6	were	be	AUX
ap-990	81	7	not	not	PART
ap-990	81	8	there	there	ADV
ap-990	81	9	,	,	PUNCT
ap-990	81	10	the	the	DET
ap-990	81	11	recogniser	recogniser	NOUN
ap-990	81	12	would	would	AUX
ap-990	81	13	take	take	VERB
ap-990	81	14	only	only	ADV
ap-990	81	15	significant	significant	ADJ
ap-990	81	16	events	event	NOUN
ap-990	81	17	in	in	ADP
ap-990	81	18	account	account	NOUN
ap-990	81	19	and	and	CCONJ
ap-990	81	20	would	would	AUX
ap-990	81	21	be	be	AUX
ap-990	81	22	able	able	ADJ
ap-990	81	23	to	to	PART
ap-990	81	24	model	model	VERB
ap-990	81	25	the	the	DET
ap-990	81	26	significant	significant	ADJ
ap-990	81	27	events	event	NOUN
ap-990	81	28	more	more	ADV
ap-990	81	29	properly	properly	ADV
ap-990	81	30	.	.	PUNCT
ap-990	82	1	this	this	PRON
ap-990	82	2	is	be	AUX
ap-990	82	3	one	one	NUM
ap-990	82	4	reason	reason	NOUN
ap-990	82	5	for	for	ADP
ap-990	82	6	using	use	VERB
ap-990	82	7	forced	force	VERB
ap-990	82	8	alignment	alignment	NOUN
ap-990	82	9	on	on	ADP
ap-990	82	10	the	the	DET
ap-990	82	11	training	training	NOUN
ap-990	82	12	dataset	dataset	NOUN
ap-990	82	13	see	see	VERB
ap-990	82	14	below	below	ADV
ap-990	82	15	.	.	PUNCT
ap-990	83	1	3	3	NUM
ap-990	83	2	previous	previous	ADJ
ap-990	83	3	tests	test	NOUN
ap-990	83	4	the	the	DET
ap-990	83	5	analysed	analyse	VERB
ap-990	83	6	databases	database	NOUN
ap-990	83	7	were	be	AUX
ap-990	83	8	used	use	VERB
ap-990	83	9	to	to	PART
ap-990	83	10	create	create	VERB
ap-990	83	11	a	a	DET
ap-990	83	12	czech	czech	ADJ
ap-990	83	13	digit	digit	NOUN
ap-990	83	14	sequence	sequence	NOUN
ap-990	83	15	recogniser	recogniser	NOUN
ap-990	83	16	that	that	PRON
ap-990	83	17	models	model	VERB
ap-990	83	18	speaker	speaker	NOUN
ap-990	83	19	non	non	ADJ
ap-990	83	20	-	-	NOUN
ap-990	83	21	speech	speech	ADJ
ap-990	83	22	events	event	NOUN
ap-990	84	1	[	[	X
ap-990	84	2	1	1	NUM
ap-990	84	3	]	]	PUNCT
ap-990	84	4	.	.	PUNCT
ap-990	85	1	in	in	ADP
ap-990	85	2	the	the	DET
ap-990	85	3	first	first	ADJ
ap-990	85	4	step	step	NOUN
ap-990	85	5	,	,	PUNCT
ap-990	85	6	only	only	ADV
ap-990	85	7	two	two	NUM
ap-990	85	8	classes	class	NOUN
ap-990	85	9	of	of	ADP
ap-990	85	10	speaker	speaker	NOUN
ap-990	85	11	non	non	ADJ
ap-990	85	12	-	-	ADJ
ap-990	85	13	speech	speech	ADJ
ap-990	85	14	events	event	NOUN
ap-990	85	15	were	be	AUX
ap-990	85	16	modelled	model	VERB
ap-990	85	17	and	and	CCONJ
ap-990	85	18	both	both	DET
ap-990	85	19	datasets	dataset	NOUN
ap-990	85	20	could	could	AUX
ap-990	85	21	be	be	AUX
ap-990	85	22	used	use	VERB
ap-990	85	23	.	.	PUNCT
ap-990	86	1	in	in	ADP
ap-990	86	2	subsequent	subsequent	ADJ
ap-990	86	3	processing	processing	NOUN
ap-990	86	4	,	,	PUNCT
ap-990	86	5	the	the	DET
ap-990	86	6	spk	spk	PROPN
ap-990	86	7	event	event	NOUN
ap-990	86	8	was	be	AUX
ap-990	86	9	divided	divide	VERB
ap-990	86	10	into	into	ADP
ap-990	86	11	separate	separate	ADJ
ap-990	86	12	plosive	plosive	ADJ
ap-990	86	13	and	and	CCONJ
ap-990	86	14	fricative	fricative	ADJ
ap-990	86	15	events	event	NOUN
ap-990	86	16	.	.	PUNCT
ap-990	87	1	therefore	therefore	ADV
ap-990	87	2	only	only	ADV
ap-990	87	3	tem	tem	PROPN
ap-990	87	4	was	be	AUX
ap-990	87	5	used	use	VERB
ap-990	87	6	for	for	ADP
ap-990	87	7	subsequent	subsequent	ADJ
ap-990	87	8	training	training	NOUN
ap-990	87	9	,	,	PUNCT
ap-990	87	10	because	because	SCONJ
ap-990	87	11	there	there	PRON
ap-990	87	12	was	be	VERB
ap-990	87	13	no	no	DET
ap-990	87	14	information	information	NOUN
ap-990	87	15	on	on	ADP
ap-990	87	16	these	these	DET
ap-990	87	17	properties	property	NOUN
ap-990	87	18	in	in	ADP
ap-990	87	19	the	the	DET
ap-990	87	20	spee	spee	NOUN
ap-990	87	21	dataset	dataset	NOUN
ap-990	87	22	.	.	PUNCT
ap-990	88	1	the	the	DET
ap-990	88	2	recogniser	recogniser	PROPN
ap-990	88	3	was	be	AUX
ap-990	88	4	tested	test	VERB
ap-990	88	5	on	on	ADP
ap-990	88	6	the	the	DET
ap-990	88	7	selection	selection	NOUN
ap-990	88	8	of	of	ADP
ap-990	88	9	both	both	DET
ap-990	88	10	datasets	dataset	NOUN
ap-990	88	11	(	(	PUNCT
ap-990	88	12	because	because	SCONJ
ap-990	88	13	of	of	ADP
ap-990	88	14	the	the	DET
ap-990	88	15	different	different	ADJ
ap-990	88	16	environment	environment	NOUN
ap-990	88	17	)	)	PUNCT
ap-990	88	18	.	.	PUNCT
ap-990	89	1	figure	figure	NOUN
ap-990	89	2	6	6	NUM
ap-990	89	3	shows	show	VERB
ap-990	89	4	the	the	DET
ap-990	89	5	recognition	recognition	NOUN
ap-990	89	6	results	result	VERB
ap-990	89	7	for	for	ADP
ap-990	89	8	the	the	DET
ap-990	89	9	testing	testing	NOUN
ap-990	89	10	data	datum	NOUN
ap-990	89	11	derived	derive	VERB
ap-990	89	12	from	from	ADP
ap-990	89	13	spee	spee	NOUN
ap-990	89	14	and	and	CCONJ
ap-990	89	15	tem	tem	PROPN
ap-990	89	16	in	in	ADP
ap-990	89	17	terms	term	NOUN
ap-990	89	18	of	of	ADP
ap-990	89	19	word	word	NOUN
ap-990	89	20	error	error	NOUN
ap-990	89	21	rate	rate	NOUN
ap-990	89	22	wer	wer	NOUN
ap-990	90	1	d	d	NOUN
ap-990	90	2	s	s	PROPN
ap-990	90	3	i	i	PRON
ap-990	90	4	n	n	PROPN
ap-990	90	5	�	�	PROPN
ap-990	90	6	�	�	PROPN
ap-990	90	7	�	�	PROPN
ap-990	90	8	�	�	PROPN
ap-990	90	9	�	�	PROPN
ap-990	90	10	�	�	PROPN
ap-990	90	11	�	�	PROPN
ap-990	90	12	�	�	PROPN
ap-990	90	13	100	100	NUM
ap-990	90	14	%	%	NOUN
ap-990	90	15	(	(	PUNCT
ap-990	90	16	1	1	NUM
ap-990	90	17	)	)	PUNCT
ap-990	90	18	where	where	SCONJ
ap-990	90	19	n	n	PRON
ap-990	90	20	means	mean	VERB
ap-990	90	21	number	number	NOUN
ap-990	90	22	of	of	ADP
ap-990	90	23	recognised	recognise	VERB
ap-990	90	24	words	word	NOUN
ap-990	90	25	,	,	PUNCT
ap-990	90	26	and	and	CCONJ
ap-990	90	27	d	d	NOUN
ap-990	90	28	stands	stand	VERB
ap-990	90	29	for	for	ADP
ap-990	90	30	deleted	delete	VERB
ap-990	90	31	words	word	NOUN
ap-990	90	32	,	,	PUNCT
ap-990	90	33	s	s	X
ap-990	90	34	for	for	ADP
ap-990	90	35	substituted	substitute	VERB
ap-990	90	36	words	word	NOUN
ap-990	90	37	and	and	CCONJ
ap-990	90	38	i	i	PRON
ap-990	90	39	for	for	ADP
ap-990	90	40	incorrectly	incorrectly	ADV
ap-990	90	41	inserted	insert	VERB
ap-990	90	42	words	word	NOUN
ap-990	90	43	.	.	PUNCT
ap-990	91	1	at	at	ADP
ap-990	91	2	the	the	DET
ap-990	91	3	beginning	beginning	NOUN
ap-990	91	4	of	of	ADP
ap-990	91	5	non	non	ADJ
ap-990	91	6	-	-	ADJ
ap-990	91	7	speech	speech	ADJ
ap-990	91	8	event	event	NOUN
ap-990	91	9	modelling	model	VERB
ap-990	91	10	there	there	PRON
ap-990	91	11	was	be	VERB
ap-990	91	12	a	a	DET
ap-990	91	13	recogniser	recogniser	NOUN
ap-990	91	14	trained	train	VERB
ap-990	91	15	on	on	ADP
ap-990	91	16	spee	spee	NOUN
ap-990	91	17	only	only	ADV
ap-990	91	18	(	(	PUNCT
ap-990	91	19	col	col	PROPN
ap-990	91	20	.	.	PROPN
ap-990	91	21	a	a	PRON
ap-990	91	22	)	)	PUNCT
ap-990	91	23	.	.	PUNCT
ap-990	92	1	in	in	ADP
ap-990	92	2	the	the	DET
ap-990	92	3	first	first	ADJ
ap-990	92	4	step	step	NOUN
ap-990	92	5	tem	tem	PROPN
ap-990	92	6	was	be	AUX
ap-990	92	7	added	add	VERB
ap-990	92	8	to	to	ADP
ap-990	92	9	the	the	DET
ap-990	92	10	training	training	NOUN
ap-990	92	11	set	set	NOUN
ap-990	92	12	and	and	CCONJ
ap-990	92	13	also	also	ADV
ap-990	92	14	two	two	NUM
ap-990	92	15	basic	basic	ADJ
ap-990	92	16	non	non	ADJ
ap-990	92	17	-	-	ADJ
ap-990	92	18	speech	speech	ADJ
ap-990	92	19	event	event	NOUN
ap-990	92	20	marks	mark	NOUN
ap-990	92	21	were	be	AUX
ap-990	92	22	added	add	VERB
ap-990	92	23	to	to	ADP
ap-990	92	24	the	the	DET
ap-990	92	25	set	set	NOUN
ap-990	92	26	of	of	ADP
ap-990	92	27	models	model	NOUN
ap-990	92	28	.	.	PUNCT
ap-990	93	1	the	the	DET
ap-990	93	2	first	first	ADJ
ap-990	93	3	retraining	retraining	NOUN
ap-990	93	4	decreased	decrease	VERB
ap-990	93	5	the	the	DET
ap-990	93	6	error	error	NOUN
ap-990	93	7	rate	rate	NOUN
ap-990	93	8	even	even	ADV
ap-990	93	9	without	without	ADP
ap-990	93	10	using	use	VERB
ap-990	93	11	non	non	ADJ
ap-990	93	12	-	-	ADJ
ap-990	93	13	speech	speech	ADJ
ap-990	93	14	events	event	NOUN
ap-990	93	15	(	(	PUNCT
ap-990	93	16	col	col	NOUN
ap-990	93	17	.	.	PROPN
ap-990	93	18	b	b	PROPN
ap-990	93	19	)	)	PUNCT
ap-990	93	20	,	,	PUNCT
ap-990	93	21	but	but	CCONJ
ap-990	93	22	the	the	DET
ap-990	93	23	recogniser	recogniser	NOUN
ap-990	93	24	that	that	PRON
ap-990	93	25	takes	take	VERB
ap-990	93	26	the	the	DET
ap-990	93	27	events	event	NOUN
ap-990	93	28	into	into	ADP
ap-990	93	29	account	account	NOUN
ap-990	93	30	has	have	VERB
ap-990	93	31	better	well	ADJ
ap-990	93	32	results	result	NOUN
ap-990	93	33	(	(	PUNCT
ap-990	93	34	col	col	PROPN
ap-990	93	35	.	.	PROPN
ap-990	93	36	c	c	NOUN
ap-990	93	37	)	)	PUNCT
ap-990	93	38	.	.	PUNCT
ap-990	94	1	additional	additional	ADJ
ap-990	94	2	retraining	retraining	NOUN
ap-990	94	3	brought	bring	VERB
ap-990	94	4	no	no	DET
ap-990	94	5	improvement	improvement	NOUN
ap-990	94	6	,	,	PUNCT
ap-990	94	7	so	so	CCONJ
ap-990	94	8	only	only	ADV
ap-990	94	9	tem	tem	PROPN
ap-990	94	10	was	be	AUX
ap-990	94	11	used	use	VERB
ap-990	94	12	for	for	ADP
ap-990	94	13	subsequent	subsequent	ADJ
ap-990	94	14	phases	phase	NOUN
ap-990	94	15	and	and	CCONJ
ap-990	94	16	the	the	DET
ap-990	94	17	general	general	ADJ
ap-990	94	18	spk	spk	PROPN
ap-990	94	19	mark	mark	PROPN
ap-990	94	20	was	be	AUX
ap-990	94	21	divided	divide	VERB
ap-990	94	22	into	into	ADP
ap-990	94	23	two	two	NUM
ap-990	94	24	classes	class	NOUN
ap-990	94	25	:	:	PUNCT
ap-990	94	26	fricative	fricative	ADJ
ap-990	94	27	(	(	PUNCT
ap-990	94	28	bre	bre	PROPN
ap-990	94	29	)	)	PUNCT
ap-990	94	30	and	and	CCONJ
ap-990	94	31	plosive	plosive	ADJ
ap-990	94	32	(	(	PUNCT
ap-990	94	33	pl	pl	NOUN
ap-990	94	34	)	)	PUNCT
ap-990	94	35	event	event	NOUN
ap-990	94	36	.	.	PUNCT
ap-990	95	1	after	after	ADP
ap-990	95	2	two	two	NUM
ap-990	95	3	retraining	retraining	ADJ
ap-990	95	4	steps	step	NOUN
ap-990	95	5	the	the	DET
ap-990	95	6	results	result	NOUN
ap-990	95	7	show	show	VERB
ap-990	95	8	that	that	SCONJ
ap-990	95	9	,	,	PUNCT
ap-990	95	10	unlike	unlike	ADP
ap-990	95	11	in	in	ADP
ap-990	95	12	the	the	DET
ap-990	95	13	case	case	NOUN
ap-990	95	14	of	of	ADP
ap-990	95	15	non	non	ADJ
ap-990	95	16	-	-	ADJ
ap-990	95	17	speech	speech	ADJ
ap-990	95	18	event	event	NOUN
ap-990	95	19	modelling	modelling	NOUN
ap-990	95	20	(	(	PUNCT
ap-990	95	21	col	col	PROPN
ap-990	95	22	.	.	PROPN
ap-990	95	23	e	e	PROPN
ap-990	95	24	,	,	PUNCT
ap-990	95	25	f	f	PROPN
ap-990	95	26	)	)	PUNCT
ap-990	95	27	,	,	PUNCT
ap-990	95	28	the	the	DET
ap-990	95	29	word	word	NOUN
ap-990	95	30	error	error	NOUN
ap-990	95	31	rate	rate	NOUN
ap-990	95	32	and	and	CCONJ
ap-990	95	33	insertion	insertion	NOUN
ap-990	95	34	error	error	NOUN
ap-990	95	35	rate	rate	NOUN
ap-990	95	36	increase	increase	VERB
ap-990	95	37	rapidly	rapidly	ADV
ap-990	95	38	without	without	ADP
ap-990	95	39	using	use	VERB
ap-990	95	40	the	the	DET
ap-990	95	41	additional	additional	ADJ
ap-990	95	42	models	model	NOUN
ap-990	95	43	(	(	PUNCT
ap-990	95	44	col	col	NOUN
ap-990	95	45	.	.	PROPN
ap-990	95	46	d	d	NOUN
ap-990	95	47	)	)	PUNCT
ap-990	95	48	.	.	PUNCT
ap-990	96	1	so	so	ADV
ap-990	96	2	the	the	DET
ap-990	96	3	experiment	experiment	NOUN
ap-990	96	4	therefore	therefore	ADV
ap-990	96	5	shows	show	VERB
ap-990	96	6	that	that	SCONJ
ap-990	96	7	the	the	DET
ap-990	96	8	non	non	ADJ
ap-990	96	9	-	-	ADJ
ap-990	96	10	speech	speech	ADJ
ap-990	96	11	event	event	NOUN
ap-990	96	12	modelling	model	VERB
ap-990	96	13	approach	approach	NOUN
ap-990	96	14	helps	help	VERB
ap-990	96	15	recognition	recognition	NOUN
ap-990	96	16	improvement	improvement	NOUN
ap-990	96	17	.	.	PUNCT
ap-990	97	1	4	4	NUM
ap-990	97	2	experiments	experiment	NOUN
ap-990	97	3	on	on	ADP
ap-990	97	4	forced	force	VERB
ap-990	97	5	-	-	PUNCT
ap-990	97	6	alignment	alignment	NOUN
ap-990	97	7	as	as	SCONJ
ap-990	97	8	noted	note	VERB
ap-990	97	9	above	above	ADV
ap-990	97	10	,	,	PUNCT
ap-990	97	11	the	the	DET
ap-990	97	12	spee	spee	PROPN
ap-990	97	13	database	database	NOUN
ap-990	97	14	divides	divide	VERB
ap-990	97	15	speaker	speaker	NOUN
ap-990	97	16	non	non	ADJ
ap-990	97	17	-	-	ADJ
ap-990	97	18	speech	speech	ADJ
ap-990	97	19	events	event	NOUN
ap-990	97	20	into	into	ADP
ap-990	97	21	two	two	NUM
ap-990	97	22	classes	class	NOUN
ap-990	97	23	only	only	ADV
ap-990	97	24	,	,	PUNCT
ap-990	97	25	but	but	CCONJ
ap-990	97	26	it	it	PRON
ap-990	97	27	is	be	AUX
ap-990	97	28	better	well	ADJ
ap-990	97	29	for	for	SCONJ
ap-990	97	30	the	the	DET
ap-990	97	31	recogniser	recogniser	NOUN
ap-990	97	32	to	to	PART
ap-990	97	33	have	have	VERB
ap-990	97	34	more	more	ADJ
ap-990	97	35	classes	class	NOUN
ap-990	97	36	of	of	ADP
ap-990	97	37	these	these	DET
ap-990	97	38	events	event	NOUN
ap-990	97	39	.	.	PUNCT
ap-990	98	1	the	the	DET
ap-990	98	2	more	more	ADV
ap-990	98	3	©	©	PROPN
ap-990	98	4	czech	czech	PROPN
ap-990	98	5	technical	technical	PROPN
ap-990	98	6	university	university	PROPN
ap-990	98	7	publishing	publishing	NOUN
ap-990	98	8	house	house	NOUN
ap-990	98	9	http://ctn.cvut.cz/ap/	http://ctn.cvut.cz/ap/	PROPN
ap-990	98	10	109	109	NUM
ap-990	98	11	acta	acta	PROPN
ap-990	98	12	polytechnica	polytechnica	PROPN
ap-990	98	13	vol	vol	NOUN
ap-990	98	14	.	.	PUNCT
ap-990	99	1	47	47	NUM
ap-990	99	2	no	no	NOUN
ap-990	99	3	.	.	PUNCT
ap-990	100	1	4–5/2007	4–5/2007	NUM
ap-990	100	2	fig	fig	NOUN
ap-990	100	3	.	.	PUNCT
ap-990	101	1	4	4	NUM
ap-990	101	2	:	:	PUNCT
ap-990	101	3	non	non	ADJ
ap-990	101	4	-	-	ADJ
ap-990	101	5	speech	speech	ADJ
ap-990	101	6	events	event	NOUN
ap-990	101	7	in	in	ADP
ap-990	101	8	the	the	DET
ap-990	101	9	whole	whole	NOUN
ap-990	101	10	(	(	PUNCT
ap-990	101	11	full	full	ADJ
ap-990	101	12	filled	fill	VERB
ap-990	101	13	bar	bar	NOUN
ap-990	101	14	)	)	PUNCT
ap-990	101	15	and	and	CCONJ
ap-990	101	16	training	training	NOUN
ap-990	101	17	(	(	PUNCT
ap-990	101	18	lined	line	VERB
ap-990	101	19	bar	bar	NOUN
ap-990	101	20	)	)	PUNCT
ap-990	101	21	tem	tem	PROPN
ap-990	101	22	dataset	dataset	VERB
ap-990	101	23	in	in	ADP
ap-990	101	24	different	different	ADJ
ap-990	101	25	noisy	noisy	ADJ
ap-990	101	26	environment	environment	NOUN
ap-990	101	27	fig	fig	NOUN
ap-990	101	28	.	.	PUNCT
ap-990	102	1	5	5	NUM
ap-990	102	2	:	:	PUNCT
ap-990	102	3	non	non	ADJ
ap-990	102	4	-	-	ADJ
ap-990	102	5	speech	speech	ADJ
ap-990	102	6	events	event	NOUN
ap-990	102	7	in	in	ADP
ap-990	102	8	the	the	DET
ap-990	102	9	whole	whole	NOUN
ap-990	102	10	(	(	PUNCT
ap-990	102	11	full	full	ADJ
ap-990	102	12	filled	fill	VERB
ap-990	102	13	bar	bar	NOUN
ap-990	102	14	)	)	PUNCT
ap-990	102	15	and	and	CCONJ
ap-990	102	16	training	training	NOUN
ap-990	102	17	(	(	PUNCT
ap-990	102	18	dashed	dash	VERB
ap-990	102	19	bar	bar	NOUN
ap-990	102	20	)	)	PUNCT
ap-990	102	21	datasets	dataset	NOUN
ap-990	102	22	for	for	ADP
ap-990	102	23	different	different	ADJ
ap-990	102	24	dialect	dialect	NOUN
ap-990	102	25	/	/	SYM
ap-990	102	26	workgroup	workgroup	NOUN
ap-990	102	27	fig	fig	NOUN
ap-990	102	28	.	.	PUNCT
ap-990	103	1	6	6	NUM
ap-990	103	2	:	:	PUNCT
ap-990	103	3	word	word	NOUN
ap-990	103	4	error	error	NOUN
ap-990	103	5	rate	rate	NOUN
ap-990	103	6	(	(	PUNCT
ap-990	103	7	dashed	dash	VERB
ap-990	103	8	bar	bar	NOUN
ap-990	103	9	)	)	PUNCT
ap-990	103	10	and	and	CCONJ
ap-990	103	11	insertion	insertion	NOUN
ap-990	103	12	error	error	NOUN
ap-990	103	13	rate	rate	NOUN
ap-990	103	14	(	(	PUNCT
ap-990	103	15	full	full	ADJ
ap-990	103	16	filled	fill	VERB
ap-990	103	17	bar	bar	NOUN
ap-990	103	18	)	)	PUNCT
ap-990	103	19	on	on	ADP
ap-990	103	20	the	the	DET
ap-990	103	21	spee	spee	NOUN
ap-990	103	22	-	-	PUNCT
ap-990	103	23	testdat	testdat	NOUN
ap-990	103	24	(	(	PUNCT
ap-990	103	25	top	top	NOUN
ap-990	103	26	)	)	PUNCT
ap-990	103	27	and	and	CCONJ
ap-990	103	28	on	on	ADP
ap-990	103	29	the	the	DET
ap-990	103	30	tem	tem	PROPN
ap-990	103	31	-	-	PUNCT
ap-990	103	32	testdat	testdat	PROPN
ap-990	103	33	(	(	PUNCT
ap-990	103	34	bottom	bottom	ADJ
ap-990	103	35	)	)	PUNCT
ap-990	103	36	classes	class	NOUN
ap-990	103	37	are	be	AUX
ap-990	103	38	used	use	VERB
ap-990	103	39	,	,	PUNCT
ap-990	103	40	the	the	PRON
ap-990	103	41	more	more	ADV
ap-990	103	42	accurately	accurately	ADV
ap-990	103	43	the	the	DET
ap-990	103	44	events	event	NOUN
ap-990	103	45	can	can	AUX
ap-990	103	46	be	be	AUX
ap-990	103	47	described	describe	VERB
ap-990	103	48	and	and	CCONJ
ap-990	103	49	modelled	model	VERB
ap-990	103	50	.	.	PUNCT
ap-990	104	1	therefore	therefore	ADV
ap-990	104	2	three	three	NUM
ap-990	104	3	models	model	NOUN
ap-990	104	4	of	of	ADP
ap-990	104	5	non	non	ADJ
ap-990	104	6	-	-	ADJ
ap-990	104	7	speech	speech	ADJ
ap-990	104	8	events	event	NOUN
ap-990	104	9	were	be	AUX
ap-990	104	10	used	use	VERB
ap-990	104	11	in	in	ADP
ap-990	104	12	the	the	DET
ap-990	104	13	recogniser	recogniser	NOUN
ap-990	104	14	above	above	ADV
ap-990	104	15	,	,	PUNCT
ap-990	104	16	which	which	PRON
ap-990	104	17	meant	mean	VERB
ap-990	104	18	using	use	VERB
ap-990	104	19	only	only	ADJ
ap-990	104	20	tem	tem	NOUN
ap-990	104	21	for	for	ADP
ap-990	104	22	subsequent	subsequent	ADJ
ap-990	104	23	retraining	retraining	NOUN
ap-990	104	24	and	and	CCONJ
ap-990	104	25	the	the	DET
ap-990	104	26	training	training	NOUN
ap-990	104	27	dataset	dataset	NOUN
ap-990	104	28	size	size	NOUN
ap-990	104	29	decreased	decrease	VERB
ap-990	104	30	.	.	PUNCT
ap-990	105	1	using	use	VERB
ap-990	105	2	the	the	DET
ap-990	105	3	spee	spee	NOUN
ap-990	105	4	database	database	NOUN
ap-990	105	5	for	for	ADP
ap-990	105	6	subsequent	subsequent	ADJ
ap-990	105	7	training	training	NOUN
ap-990	105	8	again	again	ADV
ap-990	105	9	needs	need	VERB
ap-990	105	10	a	a	DET
ap-990	105	11	decision	decision	NOUN
ap-990	105	12	on	on	ADP
ap-990	105	13	,	,	PUNCT
ap-990	105	14	whether	whether	SCONJ
ap-990	105	15	the	the	DET
ap-990	105	16	spk	spk	PROPN
ap-990	105	17	mark	mark	NOUN
ap-990	105	18	in	in	ADP
ap-990	105	19	the	the	DET
ap-990	105	20	transcription	transcription	NOUN
ap-990	105	21	closer	close	ADV
ap-990	105	22	to	to	ADP
ap-990	105	23	a	a	DET
ap-990	105	24	bre	bre	NOUN
ap-990	105	25	or	or	CCONJ
ap-990	105	26	to	to	ADP
ap-990	105	27	a	a	DET
ap-990	105	28	pl	pl	NOUN
ap-990	105	29	event	event	NOUN
ap-990	105	30	.	.	PUNCT
ap-990	106	1	the	the	DET
ap-990	106	2	htk	htk	PROPN
ap-990	106	3	system	system	NOUN
ap-990	106	4	enables	enable	VERB
ap-990	106	5	a	a	DET
ap-990	106	6	feature	feature	NOUN
ap-990	106	7	called	call	VERB
ap-990	106	8	forced	force	VERB
ap-990	106	9	alignment	alignment	NOUN
ap-990	106	10	,	,	PUNCT
ap-990	106	11	which	which	PRON
ap-990	106	12	tries	try	VERB
ap-990	106	13	to	to	PART
ap-990	106	14	recognise	recognise	VERB
ap-990	106	15	the	the	DET
ap-990	106	16	training	training	NOUN
ap-990	106	17	database	database	NOUN
ap-990	106	18	and	and	CCONJ
ap-990	106	19	puts	put	VERB
ap-990	106	20	the	the	DET
ap-990	106	21	most	most	ADV
ap-990	106	22	fitting	fitting	ADJ
ap-990	106	23	(	(	PUNCT
ap-990	106	24	probable	probable	ADJ
ap-990	106	25	)	)	PUNCT
ap-990	106	26	form	form	NOUN
ap-990	106	27	of	of	ADP
ap-990	106	28	the	the	DET
ap-990	106	29	given	give	VERB
ap-990	106	30	word	word	NOUN
ap-990	106	31	into	into	ADP
ap-990	106	32	the	the	DET
ap-990	106	33	record	record	NOUN
ap-990	106	34	.	.	PUNCT
ap-990	107	1	this	this	PRON
ap-990	107	2	is	be	AUX
ap-990	107	3	used	use	VERB
ap-990	107	4	for	for	ADP
ap-990	107	5	deciding	decide	VERB
ap-990	107	6	between	between	ADP
ap-990	107	7	pronunciation	pronunciation	NOUN
ap-990	107	8	variants	variant	NOUN
ap-990	107	9	of	of	ADP
ap-990	107	10	one	one	NUM
ap-990	107	11	word	word	NOUN
ap-990	107	12	,	,	PUNCT
ap-990	107	13	and	and	CCONJ
ap-990	107	14	in	in	ADP
ap-990	107	15	this	this	DET
ap-990	107	16	work	work	NOUN
ap-990	107	17	it	it	PRON
ap-990	107	18	was	be	AUX
ap-990	107	19	used	use	VERB
ap-990	107	20	in	in	ADP
ap-990	107	21	a	a	DET
ap-990	107	22	similar	similar	ADJ
ap-990	107	23	way	way	NOUN
ap-990	107	24	for	for	ADP
ap-990	107	25	speaker	speaker	NOUN
ap-990	107	26	non	non	ADJ
ap-990	107	27	-	-	ADJ
ap-990	107	28	speech	speech	ADJ
ap-990	107	29	events	event	NOUN
ap-990	107	30	.	.	PUNCT
ap-990	108	1	in	in	ADP
ap-990	108	2	this	this	DET
ap-990	108	3	experiment	experiment	NOUN
ap-990	108	4	the	the	DET
ap-990	108	5	spee	spee	PROPN
ap-990	108	6	dataset	dataset	NOUN
ap-990	108	7	was	be	AUX
ap-990	108	8	re	re	VERB
ap-990	108	9	-	-	VERB
ap-990	108	10	recognised	recognise	VERB
ap-990	108	11	using	use	VERB
ap-990	108	12	the	the	DET
ap-990	108	13	recogniser	recogniser	NOUN
ap-990	108	14	above	above	ADV
ap-990	108	15	.	.	PUNCT
ap-990	109	1	because	because	SCONJ
ap-990	109	2	the	the	DET
ap-990	109	3	recogniser	recogniser	NOUN
ap-990	109	4	is	be	AUX
ap-990	109	5	able	able	ADJ
ap-990	109	6	to	to	PART
ap-990	109	7	classify	classify	VERB
ap-990	109	8	the	the	DET
ap-990	109	9	spk	spk	PROPN
ap-990	109	10	event	event	NOUN
ap-990	109	11	,	,	PUNCT
ap-990	109	12	the	the	DET
ap-990	109	13	result	result	NOUN
ap-990	109	14	of	of	ADP
ap-990	109	15	this	this	DET
ap-990	109	16	recognition	recognition	NOUN
ap-990	109	17	is	be	AUX
ap-990	109	18	the	the	DET
ap-990	109	19	spee	spee	PROPN
ap-990	109	20	database	database	NOUN
ap-990	109	21	,	,	PUNCT
ap-990	109	22	which	which	PRON
ap-990	109	23	has	have	VERB
ap-990	109	24	3	3	NUM
ap-990	109	25	classes	class	NOUN
ap-990	109	26	of	of	ADP
ap-990	109	27	speaker	speaker	NOUN
ap-990	109	28	non	non	ADJ
ap-990	109	29	-	-	ADJ
ap-990	109	30	speech	speech	ADJ
ap-990	109	31	events	event	NOUN
ap-990	109	32	,	,	PUNCT
ap-990	109	33	like	like	ADP
ap-990	109	34	in	in	ADP
ap-990	109	35	the	the	DET
ap-990	109	36	tem	tem	PROPN
ap-990	109	37	database	database	NOUN
ap-990	109	38	.	.	PUNCT
ap-990	110	1	4.1	4.1	NUM
ap-990	110	2	breath	breath	NOUN
ap-990	110	3	-	-	PUNCT
ap-990	110	4	like	like	ADJ
ap-990	110	5	noise	noise	NOUN
ap-990	110	6	the	the	DET
ap-990	110	7	recogniser	recogniser	NOUN
ap-990	110	8	above	above	ADP
ap-990	110	9	uses	use	VERB
ap-990	110	10	two	two	NUM
ap-990	110	11	types	type	NOUN
ap-990	110	12	of	of	ADP
ap-990	110	13	bre	bre	PROPN
ap-990	110	14	model	model	NOUN
ap-990	110	15	.	.	PUNCT
ap-990	111	1	one	one	NUM
ap-990	111	2	is	be	AUX
ap-990	111	3	based	base	VERB
ap-990	111	4	on	on	ADP
ap-990	111	5	the	the	DET
ap-990	111	6	phonetically	phonetically	ADV
ap-990	111	7	similar	similar	ADJ
ap-990	111	8	phoneme	phoneme	NOUN
ap-990	111	9	”	"	PUNCT
ap-990	111	10	f	f	PROPN
ap-990	111	11	”	"	PUNCT
ap-990	111	12	,	,	PUNCT
ap-990	111	13	while	while	SCONJ
ap-990	111	14	the	the	DET
ap-990	111	15	second	second	NOUN
ap-990	111	16	takes	take	VERB
ap-990	111	17	advantage	advantage	NOUN
ap-990	111	18	of	of	ADP
ap-990	111	19	the	the	DET
ap-990	111	20	length	length	NOUN
ap-990	111	21	of	of	ADP
ap-990	111	22	silence	silence	NOUN
ap-990	111	23	model	model	NOUN
ap-990	111	24	.	.	PUNCT
ap-990	112	1	so	so	ADV
ap-990	112	2	it	it	PRON
ap-990	112	3	was	be	AUX
ap-990	112	4	necessary	necessary	ADJ
ap-990	112	5	to	to	PART
ap-990	112	6	decide	decide	VERB
ap-990	112	7	which	which	PRON
ap-990	112	8	should	should	AUX
ap-990	112	9	be	be	AUX
ap-990	112	10	used	use	VERB
ap-990	112	11	for	for	ADP
ap-990	112	12	subsequent	subsequent	ADJ
ap-990	112	13	processing	processing	NOUN
ap-990	112	14	.	.	PUNCT
ap-990	113	1	the	the	DET
ap-990	113	2	recognition	recognition	NOUN
ap-990	113	3	results	result	NOUN
ap-990	113	4	(	(	PUNCT
ap-990	113	5	fig	fig	NOUN
ap-990	113	6	.	.	PUNCT
ap-990	114	1	6	6	NUM
ap-990	114	2	,	,	PUNCT
ap-990	114	3	col	col	PROPN
ap-990	114	4	.	.	PROPN
ap-990	114	5	e	e	PROPN
ap-990	114	6	,	,	PUNCT
ap-990	114	7	f	f	X
ap-990	114	8	)	)	PUNCT
ap-990	114	9	show	show	VERB
ap-990	114	10	that	that	SCONJ
ap-990	114	11	in	in	ADP
ap-990	114	12	the	the	DET
ap-990	114	13	two	two	NUM
ap-990	114	14	cases	case	NOUN
ap-990	114	15	the	the	DET
ap-990	114	16	recognition	recognition	NOUN
ap-990	114	17	score	score	NOUN
ap-990	114	18	is	be	AUX
ap-990	114	19	almost	almost	ADV
ap-990	114	20	the	the	DET
ap-990	114	21	same	same	ADJ
ap-990	114	22	.	.	PUNCT
ap-990	115	1	it	it	PRON
ap-990	115	2	seems	seem	VERB
ap-990	115	3	that	that	SCONJ
ap-990	115	4	both	both	DET
ap-990	115	5	models	model	NOUN
ap-990	115	6	have	have	VERB
ap-990	115	7	the	the	DET
ap-990	115	8	same	same	ADJ
ap-990	115	9	ability	ability	NOUN
ap-990	115	10	to	to	PART
ap-990	115	11	describe	describe	VERB
ap-990	115	12	the	the	DET
ap-990	115	13	event	event	NOUN
ap-990	115	14	,	,	PUNCT
ap-990	115	15	but	but	CCONJ
ap-990	115	16	using	use	VERB
ap-990	115	17	forced	force	VERB
ap-990	115	18	alignment	alignment	NOUN
ap-990	115	19	on	on	ADP
ap-990	115	20	the	the	DET
ap-990	115	21	spee	spee	PROPN
ap-990	115	22	dataset	dataset	PROPN
ap-990	115	23	discovered	discover	VERB
ap-990	115	24	the	the	DET
ap-990	115	25	quality	quality	NOUN
ap-990	115	26	of	of	ADP
ap-990	115	27	these	these	DET
ap-990	115	28	models	model	NOUN
ap-990	115	29	.	.	PUNCT
ap-990	116	1	this	this	PRON
ap-990	116	2	was	be	AUX
ap-990	116	3	done	do	VERB
ap-990	116	4	by	by	ADP
ap-990	116	5	comparing	compare	VERB
ap-990	116	6	of	of	ADP
ap-990	116	7	the	the	DET
ap-990	116	8	results	result	NOUN
ap-990	116	9	of	of	ADP
ap-990	116	10	the	the	DET
ap-990	116	11	re	re	ADJ
ap-990	116	12	-	-	ADJ
ap-990	116	13	aligned	aligned	ADJ
ap-990	116	14	spee	spee	NOUN
ap-990	116	15	when	when	SCONJ
ap-990	116	16	using	use	VERB
ap-990	116	17	the	the	DET
ap-990	116	18	bre	bre	NOUN
ap-990	116	19	models	model	NOUN
ap-990	116	20	with	with	ADP
ap-990	116	21	a	a	DET
ap-990	116	22	different	different	ADJ
ap-990	116	23	basemodel	basemodel	NOUN
ap-990	116	24	.	.	PUNCT
ap-990	117	1	the	the	DET
ap-990	117	2	recogniser	recogniser	PROPN
ap-990	117	3	was	be	AUX
ap-990	117	4	trained	train	VERB
ap-990	117	5	in	in	ADP
ap-990	117	6	several	several	ADJ
ap-990	117	7	retraining	retraining	ADJ
ap-990	117	8	steps	step	NOUN
ap-990	117	9	.	.	PUNCT
ap-990	118	1	in	in	ADP
ap-990	118	2	the	the	DET
ap-990	118	3	last	last	ADJ
ap-990	118	4	step	step	NOUN
ap-990	118	5	(	(	PUNCT
ap-990	118	6	phase	phase	NOUN
ap-990	118	7	2	2	NUM
ap-990	118	8	)	)	PUNCT
ap-990	118	9	and	and	CCONJ
ap-990	118	10	in	in	ADP
ap-990	118	11	one	one	NUM
ap-990	118	12	preceding	precede	VERB
ap-990	118	13	step	step	NOUN
ap-990	118	14	(	(	PUNCT
ap-990	118	15	phase	phase	NOUN
ap-990	118	16	1	1	NUM
ap-990	118	17	)	)	PUNCT
ap-990	118	18	a	a	DET
ap-990	118	19	forced	force	VERB
ap-990	118	20	alignment	alignment	NOUN
ap-990	118	21	of	of	ADP
ap-990	118	22	the	the	DET
ap-990	118	23	spee	spee	PROPN
ap-990	118	24	database	database	NOUN
ap-990	118	25	was	be	AUX
ap-990	118	26	performed	perform	VERB
ap-990	118	27	table	table	NOUN
ap-990	118	28	2	2	NUM
ap-990	118	29	shows	show	VERB
ap-990	118	30	the	the	DET
ap-990	118	31	difference	difference	NOUN
ap-990	118	32	between	between	ADP
ap-990	118	33	the	the	DET
ap-990	118	34	‘	'	PUNCT
ap-990	118	35	f	f	X
ap-990	118	36	’	'	PUNCT
ap-990	118	37	-based	-based	ADJ
ap-990	118	38	and	and	CCONJ
ap-990	118	39	silence	silence	NOUN
ap-990	118	40	-	-	PUNCT
ap-990	118	41	based	base	VERB
ap-990	118	42	forced	force	VERB
ap-990	118	43	alignment	alignment	NOUN
ap-990	118	44	in	in	ADP
ap-990	118	45	those	those	DET
ap-990	118	46	two	two	NUM
ap-990	118	47	succeeding	succeed	VERB
ap-990	118	48	retraining	retrain	VERB
ap-990	118	49	phases	phase	NOUN
ap-990	118	50	.	.	PUNCT
ap-990	119	1	this	this	PRON
ap-990	119	2	led	lead	VERB
ap-990	119	3	to	to	ADP
ap-990	119	4	4	4	NUM
ap-990	119	5	comparisons	comparison	NOUN
ap-990	119	6	.	.	PUNCT
ap-990	120	1	we	we	PRON
ap-990	120	2	analysed	analyse	VERB
ap-990	120	3	whether	whether	SCONJ
ap-990	120	4	one	one	NUM
ap-990	120	5	recogniser	recogniser	NOUN
ap-990	120	6	had	have	AUX
ap-990	120	7	marked	mark	VERB
ap-990	120	8	the	the	DET
ap-990	120	9	non	non	ADJ
ap-990	120	10	-	-	ADJ
ap-990	120	11	speech	speech	ADJ
ap-990	120	12	event	event	NOUN
ap-990	120	13	with	with	ADP
ap-990	120	14	the	the	DET
ap-990	120	15	same	same	ADJ
ap-990	120	16	mark	mark	NOUN
ap-990	120	17	(	(	PUNCT
ap-990	120	18	bre	bre	PROPN
ap-990	120	19	vs	vs	ADP
ap-990	120	20	pl	pl	PROPN
ap-990	120	21	)	)	PUNCT
ap-990	120	22	as	as	ADP
ap-990	120	23	another	another	DET
ap-990	120	24	recogniser	recogniser	NOUN
ap-990	120	25	(	(	PUNCT
ap-990	120	26	substitution	substitution	NOUN
ap-990	120	27	)	)	PUNCT
ap-990	120	28	.	.	PUNCT
ap-990	121	1	and	and	CCONJ
ap-990	121	2	we	we	PRON
ap-990	121	3	noted	note	VERB
ap-990	121	4	as	as	ADP
ap-990	121	5	a	a	DET
ap-990	121	6	deletion	deletion	NOUN
ap-990	121	7	the	the	DET
ap-990	121	8	situation	situation	NOUN
ap-990	121	9	,	,	PUNCT
ap-990	121	10	when	when	SCONJ
ap-990	121	11	one	one	NUM
ap-990	121	12	of	of	ADP
ap-990	121	13	them	they	PRON
ap-990	121	14	had	have	AUX
ap-990	121	15	not	not	PART
ap-990	121	16	placed	place	VERB
ap-990	121	17	a	a	DET
ap-990	121	18	mark	mark	NOUN
ap-990	121	19	where	where	SCONJ
ap-990	121	20	the	the	DET
ap-990	121	21	second	second	ADJ
ap-990	121	22	one	one	NOUN
ap-990	121	23	marked	mark	VERB
ap-990	121	24	the	the	DET
ap-990	121	25	event	event	NOUN
ap-990	121	26	.	.	PUNCT
ap-990	122	1	in	in	ADP
ap-990	122	2	the	the	DET
ap-990	122	3	case	case	NOUN
ap-990	122	4	of	of	ADP
ap-990	122	5	’	'	PUNCT
ap-990	122	6	f	f	PROPN
ap-990	122	7	’	'	PUNCT
ap-990	122	8	-based	-based	ADJ
ap-990	122	9	bre	bre	NOUN
ap-990	122	10	models	model	NOUN
ap-990	122	11	,	,	PUNCT
ap-990	122	12	the	the	DET
ap-990	122	13	event	event	NOUN
ap-990	122	14	was	be	AUX
ap-990	122	15	marked	mark	VERB
ap-990	122	16	the	the	DET
ap-990	122	17	same	same	ADJ
ap-990	122	18	way	way	NOUN
ap-990	122	19	as	as	ADP
ap-990	122	20	by	by	ADP
ap-990	122	21	other	other	ADJ
ap-990	122	22	recognisers	recogniser	NOUN
ap-990	122	23	in	in	ADP
ap-990	122	24	89.5	89.5	NUM
ap-990	122	25	%	%	NOUN
ap-990	122	26	cases	case	NOUN
ap-990	122	27	at	at	ADP
ap-990	122	28	maximum	maximum	ADJ
ap-990	122	29	.	.	PUNCT
ap-990	123	1	this	this	PRON
ap-990	123	2	shows	show	VERB
ap-990	123	3	that	that	SCONJ
ap-990	123	4	the	the	DET
ap-990	123	5	recogniser	recogniser	NOUN
ap-990	123	6	does	do	AUX
ap-990	123	7	not	not	PART
ap-990	123	8	have	have	VERB
ap-990	123	9	stable	stable	ADJ
ap-990	123	10	models	model	NOUN
ap-990	123	11	and	and	CCONJ
ap-990	123	12	these	these	DET
ap-990	123	13	models	model	NOUN
ap-990	123	14	continue	continue	VERB
ap-990	123	15	to	to	PART
ap-990	123	16	change	change	VERB
ap-990	123	17	their	their	PRON
ap-990	123	18	properties	property	NOUN
ap-990	123	19	notably	notably	ADV
ap-990	123	20	while	while	SCONJ
ap-990	123	21	being	be	AUX
ap-990	123	22	trained	train	VERB
ap-990	123	23	.	.	PUNCT
ap-990	124	1	as	as	ADP
ap-990	124	2	a	a	DET
ap-990	124	3	result	result	NOUN
ap-990	124	4	one	one	NUM
ap-990	124	5	event	event	NOUN
ap-990	124	6	will	will	AUX
ap-990	124	7	serve	serve	VERB
ap-990	124	8	as	as	ADP
ap-990	124	9	a	a	DET
ap-990	124	10	bre	bre	NOUN
ap-990	124	11	event	event	NOUN
ap-990	124	12	for	for	ADP
ap-990	124	13	a	a	DET
ap-990	124	14	while	while	NOUN
ap-990	124	15	,	,	PUNCT
ap-990	124	16	but	but	CCONJ
ap-990	124	17	then	then	ADV
ap-990	124	18	it	it	PRON
ap-990	124	19	will	will	AUX
ap-990	124	20	be	be	AUX
ap-990	124	21	used	use	VERB
ap-990	124	22	to	to	PART
ap-990	124	23	train	train	VERB
ap-990	124	24	the	the	DET
ap-990	124	25	pl	pl	PROPN
ap-990	124	26	model	model	NOUN
ap-990	124	27	,	,	PUNCT
ap-990	124	28	because	because	SCONJ
ap-990	124	29	of	of	ADP
ap-990	124	30	the	the	DET
ap-990	124	31	non	non	NOUN
ap-990	124	32	-	-	NOUN
ap-990	124	33	stability	stability	NOUN
ap-990	124	34	of	of	ADP
ap-990	124	35	the	the	DET
ap-990	124	36	forced	force	VERB
ap-990	124	37	-	-	PUNCT
ap-990	124	38	alignment	alignment	NOUN
ap-990	124	39	results	result	NOUN
ap-990	124	40	.	.	PUNCT
ap-990	125	1	on	on	ADP
ap-990	125	2	the	the	DET
ap-990	125	3	other	other	ADJ
ap-990	125	4	hand	hand	NOUN
ap-990	125	5	,	,	PUNCT
ap-990	125	6	the	the	DET
ap-990	125	7	comparison	comparison	NOUN
ap-990	125	8	of	of	ADP
ap-990	125	9	silence	silence	NOUN
ap-990	125	10	-	-	PUNCT
ap-990	125	11	based	base	VERB
ap-990	125	12	models	model	NOUN
ap-990	125	13	in	in	ADP
ap-990	125	14	both	both	PRON
ap-990	125	15	succeeding	succeed	VERB
ap-990	125	16	phases	phase	NOUN
ap-990	125	17	shows	show	VERB
ap-990	125	18	that	that	SCONJ
ap-990	125	19	these	these	DET
ap-990	125	20	models	model	NOUN
ap-990	125	21	act	act	VERB
ap-990	125	22	in	in	ADP
ap-990	125	23	a	a	DET
ap-990	125	24	rather	rather	ADV
ap-990	125	25	stable	stable	ADJ
ap-990	125	26	way	way	NOUN
ap-990	125	27	.	.	PUNCT
ap-990	126	1	therefore	therefore	ADV
ap-990	126	2	the	the	DET
ap-990	126	3	event	event	NOUN
ap-990	126	4	will	will	AUX
ap-990	126	5	be	be	AUX
ap-990	126	6	used	use	VERB
ap-990	126	7	to	to	PART
ap-990	126	8	train	train	VERB
ap-990	126	9	only	only	ADV
ap-990	126	10	one	one	NUM
ap-990	126	11	type	type	NOUN
ap-990	126	12	of	of	ADP
ap-990	126	13	non	non	ADJ
ap-990	126	14	-	-	ADJ
ap-990	126	15	speech	speech	ADJ
ap-990	126	16	event	event	NOUN
ap-990	126	17	model	model	NOUN
ap-990	126	18	,	,	PUNCT
ap-990	126	19	and	and	CCONJ
ap-990	126	20	seems	seem	VERB
ap-990	126	21	to	to	PART
ap-990	126	22	be	be	AUX
ap-990	126	23	more	more	ADV
ap-990	126	24	suitable	suitable	ADJ
ap-990	126	25	for	for	ADP
ap-990	126	26	subsequent	subsequent	ADJ
ap-990	126	27	processing	processing	NOUN
ap-990	126	28	.	.	PUNCT
ap-990	127	1	4.2	4.2	NUM
ap-990	127	2	aligned	align	VERB
ap-990	127	3	data	data	NOUN
ap-990	127	4	-	-	PUNCT
ap-990	127	5	based	base	VERB
ap-990	127	6	recogniser	recogniser	NOUN
ap-990	127	7	the	the	DET
ap-990	127	8	first	first	ADJ
ap-990	127	9	analysis	analysis	NOUN
ap-990	127	10	(	(	PUNCT
ap-990	127	11	listening	listen	VERB
ap-990	127	12	test	test	NOUN
ap-990	127	13	)	)	PUNCT
ap-990	127	14	of	of	ADP
ap-990	127	15	the	the	DET
ap-990	127	16	re	re	ADJ
ap-990	127	17	-	-	ADJ
ap-990	127	18	aligned	aligned	ADJ
ap-990	127	19	data	datum	NOUN
ap-990	127	20	discovered	discover	VERB
ap-990	127	21	faults	fault	NOUN
ap-990	127	22	in	in	ADP
ap-990	127	23	the	the	DET
ap-990	127	24	ability	ability	NOUN
ap-990	127	25	of	of	ADP
ap-990	127	26	the	the	DET
ap-990	127	27	recogniser	recogniser	NOUN
ap-990	127	28	to	to	PART
ap-990	127	29	decide	decide	VERB
ap-990	127	30	whether	whether	SCONJ
ap-990	127	31	the	the	DET
ap-990	127	32	spk	spk	PROPN
ap-990	127	33	event	event	NOUN
ap-990	127	34	is	be	AUX
ap-990	127	35	of	of	ADP
ap-990	127	36	bre	bre	PROPN
ap-990	127	37	or	or	CCONJ
ap-990	127	38	pl	pl	PROPN
ap-990	127	39	type	type	NOUN
ap-990	127	40	.	.	PUNCT
ap-990	128	1	in	in	ADP
ap-990	128	2	some	some	DET
ap-990	128	3	cases	case	NOUN
ap-990	128	4	the	the	DET
ap-990	128	5	event	event	NOUN
ap-990	128	6	was	be	AUX
ap-990	128	7	too	too	ADV
ap-990	128	8	close	close	ADJ
ap-990	128	9	to	to	ADP
ap-990	128	10	the	the	DET
ap-990	128	11	beginning	begin	VERB
ap-990	128	12	word	word	NOUN
ap-990	128	13	,	,	PUNCT
ap-990	128	14	sometimes	sometimes	ADV
ap-990	128	15	the	the	DET
ap-990	128	16	event	event	NOUN
ap-990	128	17	was	be	AUX
ap-990	128	18	too	too	ADV
ap-990	128	19	silent	silent	ADJ
ap-990	128	20	,	,	PUNCT
ap-990	128	21	so	so	ADV
ap-990	128	22	high	high	ADJ
ap-990	128	23	accuracy	accuracy	NOUN
ap-990	128	24	could	could	AUX
ap-990	128	25	be	be	AUX
ap-990	128	26	expected	expect	VERB
ap-990	128	27	.	.	PUNCT
ap-990	129	1	however	however	ADV
ap-990	129	2	,	,	PUNCT
ap-990	129	3	the	the	DET
ap-990	129	4	results	result	NOUN
ap-990	129	5	expanded	expand	VERB
ap-990	129	6	the	the	DET
ap-990	129	7	training	training	NOUN
ap-990	129	8	database	database	NOUN
ap-990	129	9	,	,	PUNCT
ap-990	129	10	which	which	PRON
ap-990	129	11	could	could	AUX
ap-990	129	12	help	help	VERB
ap-990	129	13	the	the	DET
ap-990	129	14	recognition	recognition	NOUN
ap-990	129	15	.	.	PUNCT
ap-990	130	1	based	base	VERB
ap-990	130	2	on	on	ADP
ap-990	130	3	the	the	DET
ap-990	130	4	results	result	NOUN
ap-990	130	5	above	above	ADP
ap-990	130	6	retraining	retrain	VERB
ap-990	130	7	was	be	AUX
ap-990	130	8	performed	perform	VERB
ap-990	130	9	with	with	ADP
ap-990	130	10	the	the	DET
ap-990	130	11	use	use	NOUN
ap-990	130	12	of	of	ADP
ap-990	130	13	the	the	DET
ap-990	130	14	re	re	ADJ
ap-990	130	15	-	-	ADJ
ap-990	130	16	aligned	aligned	ADJ
ap-990	130	17	spee	spee	NOUN
ap-990	130	18	training	training	NOUN
ap-990	130	19	dataset	dataset	NOUN
ap-990	130	20	.	.	PUNCT
ap-990	131	1	both	both	DET
ap-990	131	2	kinds	kind	NOUN
ap-990	131	3	of	of	ADP
ap-990	131	4	bre	bre	PROPN
ap-990	131	5	models	model	NOUN
ap-990	131	6	were	be	AUX
ap-990	131	7	used	use	VERB
ap-990	131	8	to	to	PART
ap-990	131	9	check	check	VERB
ap-990	131	10	whether	whether	SCONJ
ap-990	131	11	the	the	DET
ap-990	131	12	comparison	comparison	NOUN
ap-990	131	13	experiment	experiment	NOUN
ap-990	131	14	has	have	VERB
ap-990	131	15	any	any	DET
ap-990	131	16	effect	effect	NOUN
ap-990	131	17	on	on	ADP
ap-990	131	18	recognition	recognition	NOUN
ap-990	131	19	accuracy	accuracy	NOUN
ap-990	131	20	.	.	PUNCT
ap-990	132	1	table	table	NOUN
ap-990	132	2	3	3	NUM
ap-990	132	3	shows	show	VERB
ap-990	132	4	that	that	SCONJ
ap-990	132	5	using	use	VERB
ap-990	132	6	a	a	DET
ap-990	132	7	different	different	ADJ
ap-990	132	8	base	base	NOUN
ap-990	132	9	model	model	NOUN
ap-990	132	10	for	for	ADP
ap-990	132	11	a	a	DET
ap-990	132	12	bre	bre	NOUN
ap-990	132	13	event	event	NOUN
ap-990	132	14	has	have	VERB
ap-990	132	15	no	no	DET
ap-990	132	16	significant	significant	ADJ
ap-990	132	17	influence	influence	NOUN
ap-990	132	18	on	on	ADP
ap-990	132	19	recognition	recognition	NOUN
ap-990	132	20	accuracy	accuracy	NOUN
ap-990	132	21	.	.	PUNCT
ap-990	133	1	after	after	ADP
ap-990	133	2	two	two	NUM
ap-990	133	3	steps	step	NOUN
ap-990	133	4	of	of	ADP
ap-990	133	5	retraining	retrain	VERB
ap-990	133	6	,	,	PUNCT
ap-990	133	7	the	the	DET
ap-990	133	8	recogniser	recogniser	NOUN
ap-990	133	9	was	be	AUX
ap-990	133	10	able	able	ADJ
ap-990	133	11	to	to	PART
ap-990	133	12	eliminate	eliminate	VERB
ap-990	133	13	all	all	DET
ap-990	133	14	insertions	insertion	NOUN
ap-990	133	15	except	except	SCONJ
ap-990	133	16	for	for	ADP
ap-990	133	17	the	the	DET
ap-990	133	18	one	one	NOUN
ap-990	133	19	that	that	PRON
ap-990	133	20	remained	remain	VERB
ap-990	133	21	.	.	PUNCT
ap-990	134	1	but	but	CCONJ
ap-990	134	2	even	even	ADV
ap-990	134	3	when	when	SCONJ
ap-990	134	4	the	the	DET
ap-990	134	5	accuracy	accuracy	NOUN
ap-990	134	6	does	do	AUX
ap-990	134	7	not	not	PART
ap-990	134	8	achieve	achieve	VERB
ap-990	134	9	the	the	DET
ap-990	134	10	value	value	NOUN
ap-990	134	11	of	of	ADP
ap-990	134	12	the	the	DET
ap-990	134	13	original	original	ADJ
ap-990	134	14	recogniser	recogniser	NOUN
ap-990	134	15	,	,	PUNCT
ap-990	134	16	the	the	DET
ap-990	134	17	re	re	NOUN
ap-990	134	18	-	-	NOUN
ap-990	134	19	alignment	alignment	NOUN
ap-990	134	20	seems	seem	VERB
ap-990	134	21	to	to	PART
ap-990	134	22	bring	bring	VERB
ap-990	134	23	an	an	DET
ap-990	134	24	improvement	improvement	NOUN
ap-990	134	25	for	for	ADP
ap-990	134	26	subsequent	subsequent	ADJ
ap-990	134	27	training	training	NOUN
ap-990	134	28	.	.	PUNCT
ap-990	135	1	4.3	4.3	NUM
ap-990	135	2	alignment	alignment	NOUN
ap-990	135	3	against	against	ADP
ap-990	135	4	listening	listen	VERB
ap-990	135	5	as	as	SCONJ
ap-990	135	6	noted	note	VERB
ap-990	135	7	above	above	ADP
ap-990	135	8	the	the	DET
ap-990	135	9	tem	tem	PROPN
ap-990	135	10	dataset	dataset	NOUN
ap-990	135	11	divides	divide	VERB
ap-990	135	12	non	non	ADJ
ap-990	135	13	-	-	ADJ
ap-990	135	14	speech	speech	ADJ
ap-990	135	15	events	event	NOUN
ap-990	135	16	into	into	ADP
ap-990	135	17	several	several	ADJ
ap-990	135	18	classes	class	NOUN
ap-990	135	19	.	.	PUNCT
ap-990	136	1	these	these	DET
ap-990	136	2	events	event	NOUN
ap-990	136	3	were	be	AUX
ap-990	136	4	marked	mark	VERB
ap-990	136	5	by	by	ADP
ap-990	136	6	the	the	DET
ap-990	136	7	human	human	ADJ
ap-990	136	8	annotator	annotator	NOUN
ap-990	136	9	,	,	PUNCT
ap-990	136	10	so	so	CCONJ
ap-990	136	11	the	the	DET
ap-990	136	12	original	original	ADJ
ap-990	136	13	transcription	transcription	NOUN
ap-990	136	14	can	can	AUX
ap-990	136	15	be	be	AUX
ap-990	136	16	used	use	VERB
ap-990	136	17	as	as	ADP
ap-990	136	18	a	a	DET
ap-990	136	19	good	good	ADJ
ap-990	136	20	basis	basis	NOUN
ap-990	136	21	for	for	ADP
ap-990	136	22	re	re	ADJ
ap-990	136	23	-	-	NOUN
ap-990	136	24	alignment	alignment	ADJ
ap-990	136	25	quality	quality	NOUN
ap-990	136	26	comparison	comparison	NOUN
ap-990	136	27	.	.	PUNCT
ap-990	137	1	this	this	DET
ap-990	137	2	transcription	transcription	NOUN
ap-990	137	3	can	can	AUX
ap-990	137	4	be	be	AUX
ap-990	137	5	considered	consider	VERB
ap-990	137	6	as	as	ADP
ap-990	137	7	a	a	DET
ap-990	137	8	good	good	ADJ
ap-990	137	9	estimate	estimate	NOUN
ap-990	137	10	of	of	ADP
ap-990	137	11	the	the	DET
ap-990	137	12	non	non	ADJ
ap-990	137	13	-	-	ADJ
ap-990	137	14	speech	speech	ADJ
ap-990	137	15	event	event	NOUN
ap-990	137	16	class	class	NOUN
ap-990	137	17	,	,	PUNCT
ap-990	137	18	and	and	CCONJ
ap-990	137	19	if	if	SCONJ
ap-990	137	20	the	the	DET
ap-990	137	21	re	re	ADJ
ap-990	137	22	-	-	NOUN
ap-990	137	23	alignment	alignment	ADJ
ap-990	137	24	phase	phase	NOUN
ap-990	137	25	marks	mark	VERB
ap-990	137	26	some	some	DET
ap-990	137	27	plosive	plosive	ADJ
ap-990	137	28	event	event	NOUN
ap-990	137	29	as	as	ADP
ap-990	137	30	bre	bre	PROPN
ap-990	137	31	in	in	ADP
ap-990	137	32	many	many	ADJ
ap-990	137	33	cases	case	NOUN
ap-990	137	34	(	(	PUNCT
ap-990	137	35	or	or	CCONJ
ap-990	137	36	vice	vice	NOUN
ap-990	137	37	versa	versa	ADV
ap-990	137	38	)	)	PUNCT
ap-990	137	39	,	,	PUNCT
ap-990	137	40	the	the	DET
ap-990	137	41	recogniser	recogniser	NOUN
ap-990	137	42	is	be	AUX
ap-990	137	43	unable	unable	ADJ
ap-990	137	44	to	to	PART
ap-990	137	45	describe	describe	VERB
ap-990	137	46	the	the	DET
ap-990	137	47	difference	difference	NOUN
ap-990	137	48	between	between	ADP
ap-990	137	49	these	these	DET
ap-990	137	50	events	event	NOUN
ap-990	137	51	and	and	CCONJ
ap-990	137	52	it	it	PRON
ap-990	137	53	needs	need	VERB
ap-990	137	54	to	to	PART
ap-990	137	55	be	be	AUX
ap-990	137	56	trained	train	VERB
ap-990	137	57	better	well	ADV
ap-990	137	58	.	.	PUNCT
ap-990	138	1	section	section	NOUN
ap-990	138	2	2.3	2.3	NUM
ap-990	138	3	shows	show	VERB
ap-990	138	4	that	that	SCONJ
ap-990	138	5	not	not	PART
ap-990	138	6	all	all	DET
ap-990	138	7	the	the	DET
ap-990	138	8	non	non	ADJ
ap-990	138	9	-	-	ADJ
ap-990	138	10	speech	speech	ADJ
ap-990	138	11	events	event	NOUN
ap-990	138	12	marked	mark	VERB
ap-990	138	13	in	in	ADP
ap-990	138	14	the	the	DET
ap-990	138	15	original	original	ADJ
ap-990	138	16	transcription	transcription	NOUN
ap-990	138	17	of	of	ADP
ap-990	138	18	the	the	DET
ap-990	138	19	speech	speech	NOUN
ap-990	138	20	datasets	dataset	NOUN
ap-990	138	21	can	can	AUX
ap-990	138	22	be	be	AUX
ap-990	138	23	considered	consider	VERB
ap-990	138	24	as	as	ADP
ap-990	138	25	a	a	DET
ap-990	138	26	suitable	suitable	ADJ
ap-990	138	27	pattern	pattern	NOUN
ap-990	138	28	for	for	ADP
ap-990	138	29	training	train	VERB
ap-990	138	30	the	the	DET
ap-990	138	31	event	event	NOUN
ap-990	138	32	model	model	NOUN
ap-990	138	33	.	.	PUNCT
ap-990	139	1	so	so	ADV
ap-990	139	2	the	the	DET
ap-990	139	3	re	re	NOUN
ap-990	139	4	-	-	NOUN
ap-990	139	5	alignment	alignment	NOUN
ap-990	139	6	can	can	AUX
ap-990	139	7	help	help	VERB
ap-990	139	8	to	to	PART
ap-990	139	9	reduce	reduce	VERB
ap-990	139	10	the	the	DET
ap-990	139	11	difference	difference	NOUN
ap-990	139	12	between	between	ADP
ap-990	139	13	the	the	DET
ap-990	139	14	marked	mark	VERB
ap-990	139	15	events	event	NOUN
ap-990	139	16	by	by	ADP
ap-990	139	17	removing	remove	VERB
ap-990	139	18	silent	silent	ADJ
ap-990	139	19	events	event	NOUN
ap-990	139	20	,	,	PUNCT
ap-990	139	21	which	which	PRON
ap-990	139	22	can	can	AUX
ap-990	139	23	be	be	AUX
ap-990	139	24	modelled	model	VERB
ap-990	139	25	by	by	ADP
ap-990	139	26	the	the	DET
ap-990	139	27	silence	silence	NOUN
ap-990	139	28	model	model	NOUN
ap-990	139	29	.	.	PUNCT
ap-990	140	1	110	110	NUM
ap-990	140	2	©	©	PROPN
ap-990	140	3	czech	czech	PROPN
ap-990	140	4	technical	technical	PROPN
ap-990	140	5	university	university	PROPN
ap-990	140	6	publishing	publishing	NOUN
ap-990	140	7	house	house	NOUN
ap-990	140	8	http://ctn.cvut.cz/ap/	http://ctn.cvut.cz/ap/	PROPN
ap-990	140	9	acta	acta	PROPN
ap-990	140	10	polytechnica	polytechnica	PROPN
ap-990	140	11	vol	vol	NOUN
ap-990	140	12	.	.	PUNCT
ap-990	141	1	47	47	NUM
ap-990	141	2	no	no	NOUN
ap-990	141	3	.	.	PUNCT
ap-990	142	1	4–5/2007	4–5/2007	NUM
ap-990	142	2	silence	silence	NOUN
ap-990	142	3	,	,	PUNCT
ap-990	142	4	phase	phase	NOUN
ap-990	142	5	1	1	NUM
ap-990	142	6	’	'	PUNCT
ap-990	142	7	f	f	NUM
ap-990	142	8	’	'	PUNCT
ap-990	142	9	,	,	PUNCT
ap-990	142	10	phase	phase	NOUN
ap-990	142	11	2	2	NUM
ap-990	142	12	‘	'	PUNCT
ap-990	142	13	f’-based	f’-based	ADJ
ap-990	142	14	,	,	PUNCT
ap-990	142	15	phase	phase	NOUN
ap-990	142	16	1	1	NUM
ap-990	142	17	subst	subst	NOUN
ap-990	142	18	.	.	PUNCT
ap-990	143	1	24.18	24.18	NUM
ap-990	143	2	%	%	NOUN
ap-990	143	3	10.44	10.44	NUM
ap-990	143	4	%	%	NOUN
ap-990	143	5	deletion	deletion	NOUN
ap-990	143	6	5.34	5.34	NUM
ap-990	143	7	%	%	NOUN
ap-990	143	8	4.14	4.14	NUM
ap-990	143	9	%	%	NOUN
ap-990	143	10	silence	silence	NOUN
ap-990	143	11	-	-	PUNCT
ap-990	143	12	based	base	VERB
ap-990	143	13	phase	phase	NOUN
ap-990	143	14	2	2	NUM
ap-990	143	15	subst	subst	NOUN
ap-990	143	16	.	.	PUNCT
ap-990	143	17	4.44	4.44	NUM
ap-990	143	18	%	%	NOUN
ap-990	143	19	18.77	18.77	NUM
ap-990	143	20	%	%	NOUN
ap-990	143	21	deletion	deletion	NOUN
ap-990	143	22	3.86	3.86	NUM
ap-990	143	23	%	%	NOUN
ap-990	143	24	7.19	7.19	NUM
ap-990	143	25	%	%	NOUN
ap-990	143	26	table	table	NOUN
ap-990	143	27	2	2	NUM
ap-990	143	28	:	:	PUNCT
ap-990	143	29	comparison	comparison	NOUN
ap-990	143	30	of	of	ADP
ap-990	143	31	re	re	ADJ
ap-990	143	32	-	-	VERB
ap-990	143	33	aligned	aligned	ADJ
ap-990	143	34	bre	bre	PROPN
ap-990	143	35	non	non	ADJ
ap-990	143	36	-	-	ADJ
ap-990	143	37	speech	speech	ADJ
ap-990	143	38	events	event	NOUN
ap-990	143	39	in	in	ADP
ap-990	143	40	different	different	ADJ
ap-990	143	41	retraining	retraining	ADJ
ap-990	143	42	phases	phase	NOUN
ap-990	143	43	acc[%	acc[%	ADP
ap-990	143	44	]	]	PUNCT
ap-990	143	45	insertions	insertion	NOUN
ap-990	143	46	’	'	PUNCT
ap-990	143	47	f’-based	f’-base	VERB
ap-990	143	48	,	,	PUNCT
ap-990	143	49	original	original	ADJ
ap-990	143	50	96.44	96.44	NUM
ap-990	143	51	3	3	NUM
ap-990	143	52	sil	sil	PROPN
ap-990	143	53	-	-	PUNCT
ap-990	143	54	based	base	VERB
ap-990	143	55	,	,	PUNCT
ap-990	143	56	original	original	ADJ
ap-990	143	57	96.44	96.44	NUM
ap-990	143	58	2	2	NUM
ap-990	143	59	’	'	PUNCT
ap-990	143	60	f’-based	f’-base	VERB
ap-990	143	61	,	,	PUNCT
ap-990	143	62	phase	phase	NOUN
ap-990	143	63	1	1	NUM
ap-990	143	64	95.96	95.96	NUM
ap-990	143	65	3	3	NUM
ap-990	143	66	sil	sil	PROPN
ap-990	143	67	-	-	PUNCT
ap-990	143	68	based	base	VERB
ap-990	143	69	,	,	PUNCT
ap-990	143	70	phase	phase	NOUN
ap-990	143	71	1	1	NUM
ap-990	143	72	95.96	95.96	NUM
ap-990	143	73	2	2	NUM
ap-990	143	74	’	'	PUNCT
ap-990	143	75	f’-based	f’-base	VERB
ap-990	143	76	,	,	PUNCT
ap-990	143	77	phase	phase	NOUN
ap-990	143	78	2	2	NUM
ap-990	143	79	96.20	96.20	NUM
ap-990	143	80	1	1	NUM
ap-990	143	81	sil	sil	PROPN
ap-990	143	82	-	-	PUNCT
ap-990	143	83	based	base	VERB
ap-990	143	84	,	,	PUNCT
ap-990	143	85	phase	phase	NOUN
ap-990	143	86	2	2	NUM
ap-990	143	87	96.32	96.32	NUM
ap-990	143	88	1	1	NUM
ap-990	143	89	table	table	NOUN
ap-990	143	90	3	3	NUM
ap-990	143	91	:	:	PUNCT
ap-990	143	92	recognition	recognition	NOUN
ap-990	143	93	results	result	NOUN
ap-990	143	94	after	after	ADP
ap-990	143	95	re	re	NOUN
ap-990	143	96	-	-	NOUN
ap-990	143	97	alignment	alignment	ADJ
ap-990	143	98	with	with	ADP
ap-990	143	99	a	a	DET
ap-990	143	100	different	different	ADJ
ap-990	143	101	bre	bre	NOUN
ap-990	143	102	basis	basis	NOUN
ap-990	143	103	table	table	NOUN
ap-990	143	104	4	4	NUM
ap-990	143	105	shows	show	VERB
ap-990	143	106	the	the	DET
ap-990	143	107	number	number	NOUN
ap-990	143	108	of	of	ADP
ap-990	143	109	non	non	ADJ
ap-990	143	110	-	-	ADJ
ap-990	143	111	speech	speech	ADJ
ap-990	143	112	event	event	NOUN
ap-990	143	113	marks	mark	VERB
ap-990	143	114	bre	bre	PROPN
ap-990	143	115	and	and	CCONJ
ap-990	143	116	pl	pl	PROPN
ap-990	143	117	that	that	PRON
ap-990	143	118	were	be	AUX
ap-990	143	119	deleted	delete	VERB
ap-990	143	120	from	from	ADP
ap-990	143	121	the	the	DET
ap-990	143	122	original	original	ADJ
ap-990	143	123	training	training	NOUN
ap-990	143	124	subset	subset	NOUN
ap-990	143	125	(	(	PUNCT
ap-990	143	126	annotated	annotate	VERB
ap-990	143	127	by	by	ADP
ap-990	143	128	a	a	DET
ap-990	143	129	human	human	ADJ
ap-990	143	130	being	being	NOUN
ap-990	143	131	)	)	PUNCT
ap-990	143	132	in	in	ADP
ap-990	143	133	different	different	ADJ
ap-990	143	134	training	training	NOUN
ap-990	143	135	phases	phase	NOUN
ap-990	143	136	and	and	CCONJ
ap-990	143	137	which	which	PRON
ap-990	143	138	were	be	AUX
ap-990	143	139	substituted	substitute	VERB
ap-990	143	140	for	for	ADP
ap-990	143	141	each	each	DET
ap-990	143	142	other	other	ADJ
ap-990	143	143	.	.	PUNCT
ap-990	144	1	for	for	SCONJ
ap-990	144	2	the	the	DET
ap-990	144	3	spee	spee	PROPN
ap-990	144	4	dataset	dataset	VERB
ap-990	144	5	there	there	PRON
ap-990	144	6	was	be	VERB
ap-990	144	7	no	no	DET
ap-990	144	8	human	human	NOUN
ap-990	144	9	-	-	PUNCT
ap-990	144	10	aligned	align	VERB
ap-990	144	11	transcription	transcription	NOUN
ap-990	144	12	for	for	ADP
ap-990	144	13	these	these	DET
ap-990	144	14	non	non	ADJ
ap-990	144	15	-	-	ADJ
ap-990	144	16	speech	speech	ADJ
ap-990	144	17	event	event	NOUN
ap-990	144	18	classes	class	NOUN
ap-990	144	19	,	,	PUNCT
ap-990	144	20	so	so	SCONJ
ap-990	144	21	there	there	PRON
ap-990	144	22	is	be	VERB
ap-990	144	23	no	no	DET
ap-990	144	24	information	information	NOUN
ap-990	144	25	about	about	ADP
ap-990	144	26	substituted	substitute	VERB
ap-990	144	27	marks	mark	NOUN
ap-990	144	28	.	.	PUNCT
ap-990	145	1	the	the	DET
ap-990	145	2	number	number	NOUN
ap-990	145	3	of	of	ADP
ap-990	145	4	deleted	delete	VERB
ap-990	145	5	marks	mark	NOUN
ap-990	145	6	of	of	ADP
ap-990	145	7	non	non	ADJ
ap-990	145	8	-	-	ADJ
ap-990	145	9	speech	speech	ADJ
ap-990	145	10	events	event	NOUN
ap-990	145	11	rises	rise	VERB
ap-990	145	12	in	in	ADP
ap-990	145	13	the	the	DET
ap-990	145	14	case	case	NOUN
ap-990	145	15	of	of	ADP
ap-990	145	16	spee	spee	PROPN
ap-990	145	17	subset	subset	NOUN
ap-990	145	18	,	,	PUNCT
ap-990	145	19	but	but	CCONJ
ap-990	145	20	the	the	DET
ap-990	145	21	difference	difference	NOUN
ap-990	145	22	between	between	ADP
ap-990	145	23	the	the	DET
ap-990	145	24	number	number	NOUN
ap-990	145	25	of	of	ADP
ap-990	145	26	deleted	delete	VERB
ap-990	145	27	marks	mark	NOUN
ap-990	145	28	in	in	ADP
ap-990	145	29	particular	particular	ADJ
ap-990	145	30	phases	phase	NOUN
ap-990	145	31	decreases	decrease	VERB
ap-990	145	32	.	.	PUNCT
ap-990	146	1	the	the	DET
ap-990	146	2	recogniser	recogniser	PROPN
ap-990	146	3	therefore	therefore	ADV
ap-990	146	4	tends	tend	VERB
ap-990	146	5	to	to	ADP
ap-990	146	6	some	some	DET
ap-990	146	7	final	final	ADJ
ap-990	146	8	form	form	NOUN
ap-990	146	9	of	of	ADP
ap-990	146	10	non	non	ADJ
ap-990	146	11	-	-	ADJ
ap-990	146	12	speech	speech	ADJ
ap-990	146	13	event	event	NOUN
ap-990	146	14	models	model	NOUN
ap-990	146	15	that	that	PRON
ap-990	146	16	consider	consider	VERB
ap-990	146	17	about	about	ADP
ap-990	146	18	2300	2300	NUM
ap-990	146	19	marks	mark	NOUN
ap-990	146	20	in	in	ADP
ap-990	146	21	the	the	DET
ap-990	146	22	original	original	ADJ
ap-990	146	23	transcription	transcription	NOUN
ap-990	146	24	as	as	ADP
ap-990	146	25	too	too	ADV
ap-990	146	26	silent	silent	ADJ
ap-990	146	27	or	or	CCONJ
ap-990	146	28	inappropriate	inappropriate	ADJ
ap-990	146	29	in	in	ADP
ap-990	146	30	some	some	DET
ap-990	146	31	other	other	ADJ
ap-990	146	32	way	way	NOUN
ap-990	146	33	.	.	PUNCT
ap-990	147	1	for	for	SCONJ
ap-990	147	2	the	the	DET
ap-990	147	3	tem	tem	PROPN
ap-990	147	4	subset	subset	VERB
ap-990	147	5	this	this	DET
ap-990	147	6	number	number	NOUN
ap-990	147	7	does	do	AUX
ap-990	147	8	not	not	PART
ap-990	147	9	change	change	VERB
ap-990	147	10	notably	notably	ADV
ap-990	147	11	.	.	PUNCT
ap-990	148	1	this	this	PRON
ap-990	148	2	may	may	AUX
ap-990	148	3	be	be	AUX
ap-990	148	4	because	because	SCONJ
ap-990	148	5	the	the	DET
ap-990	148	6	recogniser	recogniser	NOUN
ap-990	148	7	was	be	AUX
ap-990	148	8	trained	train	VERB
ap-990	148	9	on	on	ADP
ap-990	148	10	the	the	DET
ap-990	148	11	tem	tem	PROPN
ap-990	148	12	subset	subset	VERB
ap-990	148	13	in	in	ADP
ap-990	148	14	all	all	DET
ap-990	148	15	4	4	NUM
ap-990	148	16	phases	phase	NOUN
ap-990	148	17	,	,	PUNCT
ap-990	148	18	while	while	SCONJ
ap-990	148	19	the	the	DET
ap-990	148	20	spee	spee	NOUN
ap-990	148	21	was	be	AUX
ap-990	148	22	used	use	VERB
ap-990	148	23	only	only	ADV
ap-990	148	24	in	in	ADP
ap-990	148	25	the	the	DET
ap-990	148	26	last	last	ADJ
ap-990	148	27	two	two	NUM
ap-990	148	28	training	training	NOUN
ap-990	148	29	phases	phase	NOUN
ap-990	148	30	.	.	PUNCT
ap-990	149	1	the	the	DET
ap-990	149	2	substitutions	substitution	NOUN
ap-990	149	3	in	in	ADP
ap-990	149	4	table	table	NOUN
ap-990	149	5	4	4	NUM
ap-990	149	6	show	show	VERB
ap-990	149	7	that	that	SCONJ
ap-990	149	8	training	training	NOUN
ap-990	149	9	leads	lead	VERB
ap-990	149	10	to	to	ADP
ap-990	149	11	a	a	DET
ap-990	149	12	decreasing	decrease	VERB
ap-990	149	13	number	number	NOUN
ap-990	149	14	of	of	ADP
ap-990	149	15	substituted	substitute	VERB
ap-990	149	16	marks	mark	NOUN
ap-990	149	17	.	.	PUNCT
ap-990	150	1	this	this	DET
ap-990	150	2	effect	effect	NOUN
ap-990	150	3	means	mean	VERB
ap-990	150	4	that	that	SCONJ
ap-990	150	5	the	the	DET
ap-990	150	6	recogniser	recogniser	PROPN
ap-990	150	7	classifies	classify	VERB
ap-990	150	8	the	the	DET
ap-990	150	9	non	non	ADJ
ap-990	150	10	-	-	ADJ
ap-990	150	11	speech	speech	ADJ
ap-990	150	12	events	event	NOUN
ap-990	150	13	similarly	similarly	ADV
ap-990	150	14	to	to	ADP
ap-990	150	15	the	the	DET
ap-990	150	16	human	human	ADJ
ap-990	150	17	annotator	annotator	NOUN
ap-990	150	18	,	,	PUNCT
ap-990	150	19	and	and	CCONJ
ap-990	150	20	so	so	ADV
ap-990	150	21	the	the	DET
ap-990	150	22	recogniser	recogniser	NOUN
ap-990	150	23	can	can	AUX
ap-990	150	24	re	re	VERB
ap-990	150	25	-	-	VERB
ap-990	150	26	align	align	VERB
ap-990	150	27	the	the	DET
ap-990	150	28	data	datum	NOUN
ap-990	150	29	more	more	ADV
ap-990	150	30	precisely	precisely	ADV
ap-990	150	31	.	.	PUNCT
ap-990	151	1	the	the	DET
ap-990	151	2	substitutions	substitution	NOUN
ap-990	151	3	are	be	AUX
ap-990	151	4	more	more	ADV
ap-990	151	5	often	often	ADV
ap-990	151	6	caused	cause	VERB
ap-990	151	7	by	by	ADP
ap-990	151	8	marking	mark	VERB
ap-990	151	9	the	the	DET
ap-990	151	10	pl	pl	PROPN
ap-990	151	11	event	event	NOUN
ap-990	151	12	as	as	ADP
ap-990	151	13	a	a	DET
ap-990	151	14	bre	bre	NOUN
ap-990	151	15	event	event	NOUN
ap-990	151	16	.	.	PUNCT
ap-990	152	1	a	a	DET
ap-990	152	2	simple	simple	ADJ
ap-990	152	3	listening	listening	NOUN
ap-990	152	4	test	test	NOUN
ap-990	152	5	showed	show	VERB
ap-990	152	6	that	that	SCONJ
ap-990	152	7	plosive	plosive	ADJ
ap-990	152	8	non	non	ADJ
ap-990	152	9	-	-	ADJ
ap-990	152	10	speech	speech	ADJ
ap-990	152	11	events	event	NOUN
ap-990	152	12	are	be	AUX
ap-990	152	13	followed	follow	VERB
ap-990	152	14	by	by	ADP
ap-990	152	15	or	or	CCONJ
ap-990	152	16	mixed	mix	VERB
ap-990	152	17	with	with	ADP
ap-990	152	18	breath	breath	NOUN
ap-990	152	19	in	in	ADP
ap-990	152	20	some	some	DET
ap-990	152	21	cases	case	NOUN
ap-990	152	22	and	and	CCONJ
ap-990	152	23	only	only	ADV
ap-990	152	24	the	the	DET
ap-990	152	25	pl	pl	PROPN
ap-990	152	26	event	event	NOUN
ap-990	152	27	is	be	AUX
ap-990	152	28	marked	mark	VERB
ap-990	152	29	,	,	PUNCT
ap-990	152	30	so	so	CCONJ
ap-990	152	31	the	the	DET
ap-990	152	32	substitution	substitution	NOUN
ap-990	152	33	does	do	AUX
ap-990	152	34	not	not	PART
ap-990	152	35	necessarily	necessarily	ADV
ap-990	152	36	indicate	indicate	VERB
ap-990	152	37	a	a	DET
ap-990	152	38	bad	bad	ADJ
ap-990	152	39	model	model	NOUN
ap-990	152	40	of	of	ADP
ap-990	152	41	a	a	DET
ap-990	152	42	non	non	ADJ
ap-990	152	43	-	-	ADJ
ap-990	152	44	speech	speech	ADJ
ap-990	152	45	event	event	NOUN
ap-990	152	46	.	.	PUNCT
ap-990	153	1	5	5	NUM
ap-990	153	2	conclusion	conclusion	NOUN
ap-990	153	3	this	this	DET
ap-990	153	4	paper	paper	NOUN
ap-990	153	5	describes	describe	VERB
ap-990	153	6	some	some	DET
ap-990	153	7	analyses	analysis	NOUN
ap-990	153	8	and	and	CCONJ
ap-990	153	9	tests	test	NOUN
ap-990	153	10	in	in	ADP
ap-990	153	11	a	a	DET
ap-990	153	12	non	non	ADJ
ap-990	153	13	-	-	ADJ
ap-990	153	14	speech	speech	ADJ
ap-990	153	15	event	event	NOUN
ap-990	153	16	modeling	modeling	NOUN
ap-990	153	17	task	task	NOUN
ap-990	153	18	.	.	PUNCT
ap-990	154	1	the	the	DET
ap-990	154	2	analyses	analysis	NOUN
ap-990	154	3	of	of	ADP
ap-990	154	4	the	the	DET
ap-990	154	5	training	training	NOUN
ap-990	154	6	datasets	dataset	NOUN
ap-990	154	7	show	show	VERB
ap-990	154	8	some	some	DET
ap-990	154	9	properties	property	NOUN
ap-990	154	10	that	that	PRON
ap-990	154	11	can	can	AUX
ap-990	154	12	influence	influence	VERB
ap-990	154	13	recognition	recognition	NOUN
ap-990	154	14	accuracy	accuracy	NOUN
ap-990	154	15	.	.	PUNCT
ap-990	155	1	then	then	ADV
ap-990	155	2	some	some	DET
ap-990	155	3	recognition	recognition	NOUN
ap-990	155	4	tests	test	NOUN
ap-990	155	5	were	be	AUX
ap-990	155	6	performed	perform	VERB
ap-990	155	7	to	to	PART
ap-990	155	8	find	find	VERB
ap-990	155	9	the	the	DET
ap-990	155	10	best	good	ADJ
ap-990	155	11	way	way	NOUN
ap-990	155	12	to	to	AUX
ap-990	155	13	model	model	VERB
ap-990	155	14	speaker	speaker	NOUN
ap-990	155	15	non	non	ADJ
ap-990	155	16	-	-	ADJ
ap-990	155	17	speech	speech	ADJ
ap-990	155	18	events	event	NOUN
ap-990	155	19	.	.	PUNCT
ap-990	156	1	a	a	DET
ap-990	156	2	spoken	speak	VERB
ap-990	156	3	czech	czech	ADJ
ap-990	156	4	digit	digit	NOUN
ap-990	156	5	sequence	sequence	NOUN
ap-990	156	6	recogniser	recogniser	NOUN
ap-990	156	7	based	base	VERB
ap-990	156	8	on	on	ADP
ap-990	156	9	phoneme	phoneme	NOUN
ap-990	156	10	hmms	hmms	NOUN
ap-990	156	11	was	be	AUX
ap-990	156	12	used	use	VERB
ap-990	156	13	for	for	ADP
ap-990	156	14	this	this	DET
ap-990	156	15	purpose	purpose	NOUN
ap-990	156	16	.	.	PUNCT
ap-990	157	1	the	the	DET
ap-990	157	2	speech	speech	NOUN
ap-990	157	3	databases	database	NOUN
ap-990	157	4	used	use	VERB
ap-990	157	5	for	for	ADP
ap-990	157	6	the	the	DET
ap-990	157	7	experiments	experiment	NOUN
ap-990	157	8	were	be	AUX
ap-990	157	9	analysed	analyse	VERB
ap-990	157	10	,	,	PUNCT
ap-990	157	11	and	and	CCONJ
ap-990	157	12	it	it	PRON
ap-990	157	13	was	be	AUX
ap-990	157	14	found	find	VERB
ap-990	157	15	,	,	PUNCT
ap-990	157	16	that	that	SCONJ
ap-990	157	17	a	a	DET
ap-990	157	18	part	part	NOUN
ap-990	157	19	of	of	ADP
ap-990	157	20	both	both	DET
ap-990	157	21	sets	set	NOUN
ap-990	157	22	contains	contain	VERB
ap-990	157	23	a	a	DET
ap-990	157	24	notably	notably	ADV
ap-990	157	25	different	different	ADJ
ap-990	157	26	non	non	ADJ
ap-990	157	27	-	-	ADJ
ap-990	157	28	speech	speech	ADJ
ap-990	157	29	event	event	NOUN
ap-990	157	30	rate	rate	NOUN
ap-990	157	31	.	.	PUNCT
ap-990	158	1	this	this	PRON
ap-990	158	2	was	be	AUX
ap-990	158	3	caused	cause	VERB
ap-990	158	4	by	by	ADP
ap-990	158	5	the	the	DET
ap-990	158	6	different	different	ADJ
ap-990	158	7	supervision	supervision	NOUN
ap-990	158	8	in	in	ADP
ap-990	158	9	the	the	DET
ap-990	158	10	annotation	annotation	NOUN
ap-990	158	11	phase	phase	NOUN
ap-990	158	12	of	of	ADP
ap-990	158	13	database	database	NOUN
ap-990	158	14	creation	creation	NOUN
ap-990	158	15	.	.	PUNCT
ap-990	159	1	the	the	DET
ap-990	159	2	distribution	distribution	NOUN
ap-990	159	3	in	in	ADP
ap-990	159	4	different	different	ADJ
ap-990	159	5	noise	noise	NOUN
ap-990	159	6	backgrounds	background	NOUN
ap-990	159	7	supports	support	VERB
ap-990	159	8	the	the	DET
ap-990	159	9	intuitive	intuitive	ADJ
ap-990	159	10	conclusion	conclusion	NOUN
ap-990	159	11	that	that	SCONJ
ap-990	159	12	a	a	DET
ap-990	159	13	high	high	ADJ
ap-990	159	14	noise	noise	NOUN
ap-990	159	15	level	level	NOUN
ap-990	159	16	covers	cover	VERB
ap-990	159	17	non	non	ADJ
ap-990	159	18	-	-	ADJ
ap-990	159	19	speech	speech	ADJ
ap-990	159	20	events	event	NOUN
ap-990	159	21	,	,	PUNCT
ap-990	159	22	and	and	CCONJ
ap-990	159	23	so	so	ADV
ap-990	159	24	the	the	DET
ap-990	159	25	occurrence	occurrence	NOUN
ap-990	159	26	rate	rate	NOUN
ap-990	159	27	decreases	decrease	VERB
ap-990	159	28	.	.	PUNCT
ap-990	160	1	for	for	ADP
ap-990	160	2	highly	highly	ADV
ap-990	160	3	silent	silent	ADJ
ap-990	160	4	environments	environment	NOUN
ap-990	160	5	the	the	DET
ap-990	160	6	rate	rate	NOUN
ap-990	160	7	is	be	AUX
ap-990	160	8	also	also	ADV
ap-990	160	9	lower	low	ADJ
ap-990	160	10	,	,	PUNCT
ap-990	160	11	so	so	ADV
ap-990	160	12	not	not	PART
ap-990	160	13	only	only	ADV
ap-990	160	14	the	the	DET
ap-990	160	15	cleanest	clean	ADJ
ap-990	160	16	items	item	NOUN
ap-990	160	17	are	be	AUX
ap-990	160	18	best	good	ADJ
ap-990	160	19	for	for	ADP
ap-990	160	20	the	the	DET
ap-990	160	21	non	non	ADJ
ap-990	160	22	-	-	ADJ
ap-990	160	23	speech	speech	ADJ
ap-990	160	24	event	event	NOUN
ap-990	160	25	recognition	recognition	NOUN
ap-990	160	26	task	task	NOUN
ap-990	160	27	.	.	PUNCT
ap-990	161	1	the	the	DET
ap-990	161	2	analysed	analyse	VERB
ap-990	161	3	datasets	dataset	NOUN
ap-990	161	4	were	be	AUX
ap-990	161	5	used	use	VERB
ap-990	161	6	for	for	ADP
ap-990	161	7	training	train	VERB
ap-990	161	8	the	the	DET
ap-990	161	9	recogniser	recogniser	NOUN
ap-990	161	10	,	,	PUNCT
ap-990	161	11	and	and	CCONJ
ap-990	161	12	although	although	SCONJ
ap-990	161	13	they	they	PRON
ap-990	161	14	were	be	AUX
ap-990	161	15	not	not	PART
ap-990	161	16	checked	check	VERB
ap-990	161	17	to	to	PART
ap-990	161	18	ensure	ensure	VERB
ap-990	161	19	that	that	SCONJ
ap-990	161	20	they	they	PRON
ap-990	161	21	contained	contain	VERB
ap-990	161	22	only	only	ADV
ap-990	161	23	suitable	suitable	ADJ
ap-990	161	24	non	non	ADJ
ap-990	161	25	-	-	ADJ
ap-990	161	26	speech	speech	ADJ
ap-990	161	27	items	item	NOUN
ap-990	161	28	,	,	PUNCT
ap-990	161	29	using	use	VERB
ap-990	161	30	non	non	ADJ
ap-990	161	31	-	-	ADJ
ap-990	161	32	speech	speech	ADJ
ap-990	161	33	event	event	NOUN
ap-990	161	34	modeling	model	VERB
ap-990	161	35	feature	feature	NOUN
ap-990	161	36	brought	bring	VERB
ap-990	161	37	a	a	DET
ap-990	161	38	notable	notable	ADJ
ap-990	161	39	improvement	improvement	NOUN
ap-990	161	40	.	.	PUNCT
ap-990	162	1	this	this	DET
ap-990	162	2	recogniser	recogniser	PROPN
ap-990	162	3	was	be	AUX
ap-990	162	4	used	use	VERB
ap-990	162	5	to	to	PART
ap-990	162	6	re	re	VERB
ap-990	162	7	-	-	VERB
ap-990	162	8	align	align	VERB
ap-990	162	9	one	one	NUM
ap-990	162	10	of	of	ADP
ap-990	162	11	the	the	DET
ap-990	162	12	training	training	NOUN
ap-990	162	13	datasets	dataset	NOUN
ap-990	162	14	to	to	PART
ap-990	162	15	get	get	VERB
ap-990	162	16	a	a	DET
ap-990	162	17	more	more	ADV
ap-990	162	18	accurate	accurate	ADJ
ap-990	162	19	description	description	NOUN
ap-990	162	20	of	of	ADP
ap-990	162	21	the	the	DET
ap-990	162	22	non	non	ADJ
ap-990	162	23	-	-	ADJ
ap-990	162	24	speech	speech	ADJ
ap-990	162	25	events	event	NOUN
ap-990	162	26	.	.	PUNCT
ap-990	163	1	this	this	PRON
ap-990	163	2	reduced	reduce	VERB
ap-990	163	3	the	the	DET
ap-990	163	4	insertion	insertion	NOUN
ap-990	163	5	error	error	NOUN
ap-990	163	6	.	.	PUNCT
ap-990	164	1	the	the	DET
ap-990	164	2	choice	choice	NOUN
ap-990	164	3	of	of	ADP
ap-990	164	4	the	the	DET
ap-990	164	5	model	model	NOUN
ap-990	164	6	that	that	PRON
ap-990	164	7	stands	stand	VERB
ap-990	164	8	as	as	ADP
ap-990	164	9	a	a	DET
ap-990	164	10	basis	basis	NOUN
ap-990	164	11	for	for	ADP
ap-990	164	12	non	non	ADJ
ap-990	164	13	-	-	ADJ
ap-990	164	14	speech	speech	ADJ
ap-990	164	15	events	event	NOUN
ap-990	164	16	seems	seem	VERB
ap-990	164	17	to	to	PART
ap-990	164	18	be	be	AUX
ap-990	164	19	of	of	ADP
ap-990	164	20	less	less	ADJ
ap-990	164	21	importance	importance	NOUN
ap-990	164	22	,	,	PUNCT
ap-990	164	23	because	because	SCONJ
ap-990	164	24	after	after	SCONJ
ap-990	164	25	the	the	DET
ap-990	164	26	retraining	retraining	ADJ
ap-990	164	27	difference	difference	NOUN
ap-990	164	28	slowly	slowly	ADV
ap-990	164	29	disappears	disappear	VERB
ap-990	164	30	.	.	PUNCT
ap-990	165	1	acknowledgments	acknowledgment	NOUN
ap-990	165	2	the	the	DET
ap-990	165	3	work	work	NOUN
ap-990	165	4	presented	present	VERB
ap-990	165	5	here	here	ADV
ap-990	165	6	was	be	AUX
ap-990	165	7	supported	support	VERB
ap-990	165	8	by	by	ADP
ap-990	165	9	gačr	gačr	NOUN
ap-990	165	10	102/05/0278	102/05/0278	NUM
ap-990	165	11	“	"	PUNCT
ap-990	165	12	new	new	ADJ
ap-990	165	13	trends	trend	NOUN
ap-990	165	14	in	in	ADP
ap-990	165	15	voice	voice	NOUN
ap-990	165	16	technologies	technology	NOUN
ap-990	165	17	research	research	NOUN
ap-990	165	18	and	and	CCONJ
ap-990	165	19	usage	usage	NOUN
ap-990	165	20	”	"	PUNCT
ap-990	165	21	,	,	PUNCT
ap-990	165	22	avčr	avčr	NOUN
ap-990	165	23	1et201210402	1et201210402	NUM
ap-990	166	1	“	"	PUNCT
ap-990	166	2	voice	voice	NOUN
ap-990	166	3	technologies	technology	NOUN
ap-990	166	4	in	in	ADP
ap-990	166	5	information	information	NOUN
ap-990	166	6	systems	system	NOUN
ap-990	166	7	”	"	PUNCT
ap-990	166	8	,	,	PUNCT
ap-990	166	9	iga	iga	PROPN
ap-990	166	10	mz	mz	PROPN
ap-990	166	11	čr	čr	PROPN
ap-990	166	12	nr8287	nr8287	PROPN
ap-990	166	13	-	-	PUNCT
ap-990	166	14	3/2005	3/2005	NUM
ap-990	166	15	and	and	CCONJ
ap-990	166	16	research	research	NOUN
ap-990	166	17	activity	activity	NOUN
ap-990	166	18	msm	msm	NOUN
ap-990	166	19	6840770014	6840770014	NUM
ap-990	166	20	”	"	PUNCT
ap-990	166	21	research	research	NOUN
ap-990	166	22	in	in	ADP
ap-990	166	23	the	the	DET
ap-990	166	24	area	area	NOUN
ap-990	166	25	of	of	ADP
ap-990	166	26	the	the	DET
ap-990	166	27	prospective	prospective	ADJ
ap-990	166	28	information	information	NOUN
ap-990	166	29	and	and	CCONJ
ap-990	166	30	navigation	navigation	NOUN
ap-990	166	31	technologies	technology	NOUN
ap-990	166	32	”	"	PUNCT
ap-990	166	33	.	.	PUNCT
ap-990	167	1	references	reference	NOUN
ap-990	167	2	[	[	X
ap-990	167	3	1	1	NUM
ap-990	167	4	]	]	PUNCT
ap-990	167	5	rajnoha	rajnoha	PROPN
ap-990	167	6	,	,	PUNCT
ap-990	167	7	j.	j.	PROPN
ap-990	167	8	:	:	PUNCT
ap-990	167	9	modeling	modeling	NOUN
ap-990	167	10	of	of	ADP
ap-990	167	11	speaker	speaker	NOUN
ap-990	167	12	non	non	ADJ
ap-990	167	13	-	-	NOUN
ap-990	167	14	speech	speech	ADJ
ap-990	167	15	events	event	NOUN
ap-990	167	16	in	in	ADP
ap-990	167	17	robust	robust	ADJ
ap-990	167	18	speech	speech	NOUN
ap-990	167	19	recognition	recognition	NOUN
ap-990	167	20	.	.	PUNCT
ap-990	168	1	proceedings	proceeding	NOUN
ap-990	168	2	of	of	ADP
ap-990	168	3	the	the	DET
ap-990	168	4	16th	16th	ADJ
ap-990	168	5	czech	czech	ADJ
ap-990	168	6	-	-	PUNCT
ap-990	168	7	german	german	ADJ
ap-990	168	8	workshop	workshop	NOUN
ap-990	168	9	on	on	ADP
ap-990	168	10	speech	speech	NOUN
ap-990	168	11	processing	processing	NOUN
ap-990	168	12	,	,	PUNCT
ap-990	168	13	prague	prague	NOUN
ap-990	168	14	:	:	PUNCT
ap-990	168	15	academy	academy	PROPN
ap-990	168	16	of	of	ADP
ap-990	168	17	sciences	sciences	PROPN
ap-990	168	18	of	of	ADP
ap-990	168	19	the	the	DET
ap-990	168	20	czech	czech	PROPN
ap-990	168	21	republic	republic	NOUN
ap-990	168	22	,	,	PUNCT
ap-990	168	23	institute	institute	NOUN
ap-990	168	24	of	of	ADP
ap-990	168	25	radioengineering	radioengineering	NOUN
ap-990	168	26	and	and	CCONJ
ap-990	168	27	electronics	electronic	NOUN
ap-990	168	28	,	,	PUNCT
ap-990	168	29	2006	2006	NUM
ap-990	168	30	,	,	PUNCT
ap-990	168	31	p.	p.	NOUN
ap-990	168	32	149–155	149–155	NUM
ap-990	168	33	.	.	PUNCT
ap-990	169	1	[	[	X
ap-990	169	2	2	2	NUM
ap-990	169	3	]	]	X
ap-990	169	4	young	young	ADJ
ap-990	169	5	,	,	PUNCT
ap-990	169	6	s.	s.	PROPN
ap-990	169	7	et	et	PROPN
ap-990	169	8	al	al	PROPN
ap-990	169	9	.	.	PROPN
ap-990	169	10	:	:	PUNCT
ap-990	170	1	the	the	DET
ap-990	170	2	htk	htk	PROPN
ap-990	170	3	book	book	NOUN
ap-990	170	4	(	(	PUNCT
ap-990	170	5	for	for	ADP
ap-990	170	6	htk	htk	PROPN
ap-990	170	7	version	version	PROPN
ap-990	170	8	3.2.1	3.2.1	NUM
ap-990	170	9	)	)	PUNCT
ap-990	170	10	cambridge	cambridge	PROPN
ap-990	170	11	university	university	PROPN
ap-990	170	12	engineering	engineering	PROPN
ap-990	170	13	department	department	PROPN
ap-990	170	14	,	,	PUNCT
ap-990	170	15	2002	2002	NUM
ap-990	170	16	.	.	PUNCT
ap-990	171	1	[	[	X
ap-990	171	2	3	3	NUM
ap-990	171	3	]	]	X
ap-990	171	4	gajic	gajic	PROPN
ap-990	171	5	,	,	PUNCT
ap-990	171	6	b.	b.	PROPN
ap-990	171	7	,	,	PUNCT
ap-990	171	8	markhus	markhus	PROPN
ap-990	171	9	,	,	PUNCT
ap-990	171	10	v.	v.	ADV
ap-990	171	11	,	,	PUNCT
ap-990	171	12	pettersen	pettersen	NOUN
ap-990	171	13	,	,	PUNCT
ap-990	171	14	s.	s.	PROPN
ap-990	171	15	g.	g.	PROPN
ap-990	171	16	,	,	PUNCT
ap-990	171	17	johnsen	johnsen	PROPN
ap-990	171	18	,	,	PUNCT
ap-990	171	19	m.	m.	PROPN
ap-990	171	20	h.	h.	PROPN
ap-990	171	21	:	:	PUNCT
ap-990	171	22	automatic	automatic	ADJ
ap-990	171	23	recognition	recognition	NOUN
ap-990	171	24	of	of	ADP
ap-990	171	25	spontaneously	spontaneously	ADV
ap-990	171	26	dictated	dictate	VERB
ap-990	171	27	medical	medical	ADJ
ap-990	171	28	records	record	NOUN
ap-990	171	29	for	for	ADP
ap-990	171	30	norwegian	norwegian	PROPN
ap-990	171	31	.	.	PUNCT
ap-990	172	1	cost278	cost278	PROPN
ap-990	172	2	and	and	CCONJ
ap-990	172	3	isca	isca	PROPN
ap-990	172	4	tutorial	tutorial	NOUN
ap-990	172	5	and	and	CCONJ
ap-990	172	6	research	research	NOUN
ap-990	172	7	workshop	workshop	NOUN
ap-990	172	8	–	–	PUNCT
ap-990	172	9	robust	robust	ADJ
ap-990	172	10	2004	2004	NUM
ap-990	172	11	,	,	PUNCT
ap-990	172	12	2004	2004	NUM
ap-990	172	13	.	.	PUNCT
ap-990	173	1	[	[	X
ap-990	173	2	4	4	NUM
ap-990	173	3	]	]	X
ap-990	173	4	shriberg	shriberg	NOUN
ap-990	173	5	,	,	PUNCT
ap-990	173	6	e.	e.	PROPN
ap-990	173	7	e.	e.	PROPN
ap-990	173	8	:	:	PUNCT
ap-990	173	9	phonetic	phonetic	ADJ
ap-990	173	10	consequences	consequence	NOUN
ap-990	173	11	of	of	ADP
ap-990	173	12	speech	speech	NOUN
ap-990	173	13	disfluency	disfluency	NOUN
ap-990	173	14	.	.	PUNCT
ap-990	174	1	proceedings	proceeding	NOUN
ap-990	174	2	of	of	ADP
ap-990	174	3	the	the	DET
ap-990	174	4	international	international	ADJ
ap-990	174	5	congress	congress	PROPN
ap-990	174	6	of	of	ADP
ap-990	174	7	phonetic	phonetic	ADJ
ap-990	174	8	sciences	sciences	PROPN
ap-990	174	9	,	,	PUNCT
ap-990	174	10	san	san	PROPN
ap-990	174	11	francisco	francisco	PROPN
ap-990	174	12	,	,	PUNCT
ap-990	174	13	1999	1999	NUM
ap-990	174	14	,	,	PUNCT
ap-990	174	15	p.	p.	NOUN
ap-990	174	16	619–622	619–622	NUM
ap-990	174	17	.	.	PUNCT
ap-990	175	1	[	[	X
ap-990	175	2	5	5	X
ap-990	175	3	]	]	X
ap-990	175	4	speecon	speecon	NOUN
ap-990	175	5	project	project	NOUN
ap-990	175	6	webpage	webpage	NOUN
ap-990	175	7	.	.	PUNCT
ap-990	176	1	http://www.speechdat.org/speecon	http://www.speechdat.org/speecon	PROPN
ap-990	176	2	.	.	PUNCT
ap-990	177	1	josef	josef	PROPN
ap-990	177	2	rajnoha	rajnoha	PROPN
ap-990	177	3	e	e	PROPN
ap-990	178	1	-	-	NOUN
ap-990	178	2	mail	mail	NOUN
ap-990	178	3	:	:	PUNCT
ap-990	178	4	rajnoj1@fel.cvut.cz	rajnoj1@fel.cvut.cz	X
ap-990	178	5	dept	dept	NOUN
ap-990	178	6	.	.	PROPN
ap-990	178	7	of	of	ADP
ap-990	178	8	circuit	circuit	NOUN
ap-990	178	9	theory	theory	NOUN
ap-990	178	10	czech	czech	PROPN
ap-990	178	11	technical	technical	PROPN
ap-990	178	12	university	university	PROPN
ap-990	178	13	in	in	ADP
ap-990	178	14	prague	prague	PROPN
ap-990	178	15	faculty	faculty	NOUN
ap-990	178	16	of	of	ADP
ap-990	178	17	electrical	electrical	ADJ
ap-990	178	18	engineering	engineering	NOUN
ap-990	178	19	technická	technická	NOUN
ap-990	178	20	2	2	NUM
ap-990	178	21	166	166	NUM
ap-990	178	22	27	27	NUM
ap-990	178	23	praha	praha	NOUN
ap-990	178	24	,	,	PUNCT
ap-990	178	25	czech	czech	PROPN
ap-990	178	26	republic	republic	NOUN
ap-990	178	27	©	©	PROPN
ap-990	178	28	czech	czech	PROPN
ap-990	178	29	technical	technical	PROPN
ap-990	178	30	university	university	PROPN
ap-990	178	31	publishing	publishing	NOUN
ap-990	178	32	house	house	NOUN
ap-990	178	33	http://ctn.cvut.cz/ap/	http://ctn.cvut.cz/ap/	PROPN
ap-990	178	34	111	111	NUM
ap-990	178	35	acta	acta	PROPN
ap-990	178	36	polytechnica	polytechnica	PROPN
ap-990	178	37	vol	vol	NOUN
ap-990	178	38	.	.	PUNCT
ap-990	179	1	47	47	NUM
ap-990	179	2	no	no	NOUN
ap-990	179	3	.	.	PUNCT
ap-990	180	1	4–5/2007	4–5/2007	NUM
ap-990	180	2	phase	phase	NOUN
ap-990	180	3	deleted	delete	VERB
ap-990	180	4	substituted	substitute	VERB
ap-990	180	5	speecon	speecon	NOUN
ap-990	180	6	phase	phase	NOUN
ap-990	180	7	1	1	NUM
ap-990	180	8	before	before	ADP
ap-990	180	9	re	re	NOUN
ap-990	180	10	-	-	VERB
ap-990	180	11	align	align	ADJ
ap-990	180	12	1116	1116	NUM
ap-990	180	13	–	–	PUNCT
ap-990	180	14	phase	phase	NOUN
ap-990	180	15	2	2	NUM
ap-990	180	16	before	before	ADP
ap-990	180	17	re	re	NOUN
ap-990	180	18	-	-	VERB
ap-990	180	19	align	align	ADJ
ap-990	180	20	1723	1723	NUM
ap-990	180	21	–	–	PUNCT
ap-990	180	22	phase	phase	NOUN
ap-990	180	23	1	1	NUM
ap-990	180	24	after	after	ADP
ap-990	180	25	re	re	NOUN
ap-990	180	26	-	-	VERB
ap-990	180	27	align	align	ADJ
ap-990	180	28	2169	2169	NUM
ap-990	180	29	–	–	PUNCT
ap-990	180	30	phase	phase	NOUN
ap-990	180	31	2	2	NUM
ap-990	180	32	after	after	ADP
ap-990	180	33	re	re	NOUN
ap-990	180	34	-	-	VERB
ap-990	180	35	align	align	ADJ
ap-990	180	36	2282	2282	NUM
ap-990	180	37	–	–	PUNCT
ap-990	180	38	temic	temic	NOUN
ap-990	180	39	phase	phase	NOUN
ap-990	180	40	1	1	NUM
ap-990	180	41	before	before	ADP
ap-990	180	42	re	re	NOUN
ap-990	180	43	-	-	VERB
ap-990	180	44	align	align	ADJ
ap-990	180	45	1381	1381	NUM
ap-990	180	46	424	424	NUM
ap-990	180	47	phase	phase	NOUN
ap-990	180	48	2	2	NUM
ap-990	180	49	before	before	ADP
ap-990	180	50	re	re	NOUN
ap-990	180	51	-	-	VERB
ap-990	180	52	align	align	ADJ
ap-990	180	53	1381	1381	NUM
ap-990	180	54	566	566	NUM
ap-990	180	55	phase	phase	NOUN
ap-990	180	56	1	1	NUM
ap-990	180	57	after	after	ADP
ap-990	180	58	re	re	NOUN
ap-990	180	59	-	-	VERB
ap-990	180	60	align	align	ADJ
ap-990	180	61	1296	1296	NUM
ap-990	180	62	458	458	NUM
ap-990	180	63	phase	phase	NOUN
ap-990	180	64	2	2	NUM
ap-990	180	65	after	after	ADP
ap-990	180	66	re	re	NOUN
ap-990	180	67	-	-	VERB
ap-990	180	68	align	align	ADJ
ap-990	180	69	1286	1286	NUM
ap-990	180	70	448	448	NUM
ap-990	180	71	table	table	NOUN
ap-990	180	72	4	4	NUM
ap-990	180	73	:	:	PUNCT
ap-990	180	74	comparison	comparison	NOUN
ap-990	180	75	of	of	ADP
ap-990	180	76	non	non	ADJ
ap-990	180	77	-	-	ADJ
ap-990	180	78	speech	speech	ADJ
ap-990	180	79	events	event	NOUN
ap-990	180	80	marked	mark	VERB
ap-990	180	81	in	in	ADP
ap-990	180	82	the	the	DET
ap-990	180	83	original	original	ADJ
ap-990	180	84	and	and	CCONJ
ap-990	180	85	re	re	ADJ
ap-990	180	86	-	-	ADJ
ap-990	180	87	aligned	aligned	ADJ
ap-990	180	88	training	training	NOUN
ap-990	180	89	subsets	subset	NOUN
ap-990	180	90	table	table	NOUN
ap-990	180	91	of	of	ADP
ap-990	180	92	contents	content	NOUN
ap-990	180	93	detection	detection	NOUN
ap-990	180	94	of	of	ADP
ap-990	180	95	facial	facial	ADJ
ap-990	180	96	features	feature	NOUN
ap-990	180	97	in	in	ADP
ap-990	180	98	scale	scale	NOUN
ap-990	180	99	-	-	PUNCT
ap-990	180	100	space	space	NOUN
ap-990	180	101	3	3	NUM
ap-990	180	102	p.	p.	NOUN
ap-990	180	103	hosten	hosten	NOUN
ap-990	180	104	,	,	PUNCT
ap-990	180	105	m.	m.	NOUN
ap-990	180	106	asbach	asbach	NOUN
ap-990	180	107	valuating	valuate	VERB
ap-990	180	108	the	the	DET
ap-990	180	109	investment	investment	NOUN
ap-990	180	110	efficiency	efficiency	NOUN
ap-990	180	111	of	of	ADP
ap-990	180	112	distribution	distribution	NOUN
ap-990	180	113	companies	company	NOUN
ap-990	180	114	8	8	NUM
ap-990	180	115	m.	m.	NOUN
ap-990	180	116	karajica	karajica	PROPN
ap-990	180	117	valuation	valuation	PROPN
ap-990	180	118	of	of	ADP
ap-990	180	119	companies	company	NOUN
ap-990	180	120	14	14	NUM
ap-990	180	121	jiøí	jiøí	NOUN
ap-990	180	122	lisník	lisník	VERB
ap-990	180	123	an	an	DET
ap-990	180	124	improved	improved	ADJ
ap-990	180	125	version	version	NOUN
ap-990	180	126	of	of	ADP
ap-990	180	127	the	the	DET
ap-990	180	128	fluxgate	fluxgate	ADJ
ap-990	180	129	compass	compass	NOUN
ap-990	180	130	module	module	NOUN
ap-990	180	131	18	18	NUM
ap-990	180	132	v.	v.	ADP
ap-990	180	133	petrucha	petrucha	PROPN
ap-990	180	134	robust	robust	ADJ
ap-990	180	135	detection	detection	NOUN
ap-990	180	136	of	of	ADP
ap-990	180	137	point	point	NOUN
ap-990	180	138	correspondences	correspondence	NOUN
ap-990	180	139	in	in	ADP
ap-990	180	140	stereo	stereo	ADJ
ap-990	180	141	images	image	NOUN
ap-990	180	142	23	23	NUM
ap-990	180	143	a.	a.	NOUN
ap-990	180	144	stojanovic	stojanovic	PROPN
ap-990	180	145	,	,	PUNCT
ap-990	180	146	m.	m.	NOUN
ap-990	180	147	unger	unger	NOUN
ap-990	180	148	assessment	assessment	NOUN
ap-990	180	149	of	of	ADP
ap-990	180	150	human	human	ADJ
ap-990	180	151	hemodynamics	hemodynamic	NOUN
ap-990	180	152	under	under	ADP
ap-990	180	153	hyperand	hyperand	NOUN
ap-990	180	154	microgravity	microgravity	NOUN
ap-990	180	155	:	:	PUNCT
ap-990	180	156	results	result	NOUN
ap-990	180	157	of	of	ADP
ap-990	180	158	two	two	NUM
ap-990	180	159	aachen	aachen	PROPN
ap-990	180	160	university	university	PROPN
ap-990	180	161	parabolic	parabolic	ADJ
ap-990	180	162	flight	flight	NOUN
ap-990	180	163	experiments	experiment	NOUN
ap-990	180	164	29	29	NUM
ap-990	180	165	n.	n.	PROPN
ap-990	180	166	blanik	blanik	PROPN
ap-990	180	167	,	,	PUNCT
ap-990	180	168	m.	m.	NOUN
ap-990	180	169	hülsbusch	hülsbusch	PROPN
ap-990	180	170	,	,	PUNCT
ap-990	180	171	m.	m.	NOUN
ap-990	180	172	herzog	herzog	PROPN
ap-990	180	173	,	,	PUNCT
ap-990	180	174	c.	c.	PROPN
ap-990	180	175	r.	r.	PROPN
ap-990	180	176	blazek	blazek	PROPN
ap-990	180	177	refactorisation	refactorisation	NOUN
ap-990	180	178	methods	method	NOUN
ap-990	180	179	for	for	ADP
ap-990	180	180	ttcn-3	ttcn-3	NUM
ap-990	180	181	33	33	NUM
ap-990	180	182	l.	l.	PROPN
ap-990	180	183	eros	eros	PROPN
ap-990	180	184	,	,	PUNCT
ap-990	180	185	f.	f.	PROPN
ap-990	180	186	bozoki	bozoki	PROPN
ap-990	180	187	extending	extend	VERB
ap-990	180	188	the	the	DET
ap-990	180	189	life	life	NOUN
ap-990	180	190	time	time	NOUN
ap-990	180	191	of	of	ADP
ap-990	180	192	a	a	DET
ap-990	180	193	nuclear	nuclear	ADJ
ap-990	180	194	power	power	NOUN
ap-990	180	195	plant	plant	NOUN
ap-990	180	196	:	:	PUNCT
ap-990	180	197	impact	impact	NOUN
ap-990	180	198	on	on	ADP
ap-990	180	199	nuclear	nuclear	ADJ
ap-990	180	200	liabilities	liability	NOUN
ap-990	180	201	in	in	ADP
ap-990	180	202	the	the	DET
ap-990	180	203	czech	czech	PROPN
ap-990	180	204	republic	republic	NOUN
ap-990	180	205	38	38	NUM
ap-990	180	206	l.	l.	PROPN
ap-990	180	207	havlíèek	havlíèek	NOUN
ap-990	180	208	negative	negative	ADJ
ap-990	180	209	chromatic	chromatic	ADJ
ap-990	180	210	dispersion	dispersion	NOUN
ap-990	180	211	generated	generate	VERB
ap-990	180	212	by	by	ADP
ap-990	180	213	introducing	introduce	VERB
ap-990	180	214	curvature	curvature	NOUN
ap-990	180	215	into	into	ADP
ap-990	180	216	photonic	photonic	ADJ
ap-990	180	217	crystal	crystal	NOUN
ap-990	180	218	fiber	fiber	NOUN
ap-990	180	219	43	43	NUM
ap-990	180	220	m.	m.	NOUN
ap-990	180	221	lucki	lucki	PROPN
ap-990	180	222	developing	develop	VERB
ap-990	180	223	e	e	NOUN
ap-990	180	224	-	-	ADJ
ap-990	180	225	learning	learn	VERB
ap-990	180	226	courses	course	NOUN
ap-990	180	227	for	for	ADP
ap-990	180	228	mobile	mobile	ADJ
ap-990	180	229	devices	device	NOUN
ap-990	180	230	48	48	NUM
ap-990	180	231	r.	r.	PROPN
ap-990	180	232	szabados	szabados	PROPN
ap-990	180	233	,	,	PUNCT
ap-990	180	234	k.	k.	PROPN
ap-990	180	235	sipos	sipos	PROPN
ap-990	181	1	the	the	DET
ap-990	181	2	influence	influence	NOUN
ap-990	181	3	of	of	ADP
ap-990	181	4	pzt	pzt	NOUN
ap-990	181	5	actuators	actuator	NOUN
ap-990	181	6	positioning	position	VERB
ap-990	181	7	in	in	ADP
ap-990	181	8	active	active	ADJ
ap-990	181	9	structural	structural	ADJ
ap-990	181	10	acoustic	acoustic	ADJ
ap-990	181	11	control	control	NOUN
ap-990	181	12	55	55	NUM
ap-990	181	13	p.	p.	NOUN
ap-990	181	14	švec	švec	NOUN
ap-990	181	15	,	,	PUNCT
ap-990	181	16	v.	v.	ADP
ap-990	181	17	jandák	jandák	PROPN
ap-990	181	18	computer	computer	NOUN
ap-990	181	19	controlled	control	VERB
ap-990	181	20	switching	switch	VERB
ap-990	181	21	device	device	NOUN
ap-990	181	22	for	for	ADP
ap-990	181	23	deep	deep	ADJ
ap-990	181	24	brain	brain	NOUN
ap-990	181	25	stimulation	stimulation	NOUN
ap-990	181	26	59	59	NUM
ap-990	181	27	j.	j.	PROPN
ap-990	181	28	tauchmanová	tauchmanová	AUX
ap-990	181	29	modeling	model	VERB
ap-990	181	30	measurement	measurement	NOUN
ap-990	181	31	uncertainty	uncertainty	NOUN
ap-990	181	32	in	in	ADP
ap-990	181	33	room	room	NOUN
ap-990	181	34	acoustics	acoustic	NOUN
ap-990	181	35	63	63	NUM
ap-990	181	36	p.	p.	NOUN
ap-990	181	37	dietrich	dietrich	PROPN
ap-990	181	38	wireless	wireless	PROPN
ap-990	181	39	and	and	CCONJ
ap-990	181	40	non	non	ADJ
ap-990	181	41	-	-	ADJ
ap-990	181	42	contact	contact	ADJ
ap-990	181	43	ecg	ecg	PROPN
ap-990	181	44	measurement	measurement	NOUN
ap-990	181	45	system	system	NOUN
ap-990	181	46	–	–	PUNCT
ap-990	181	47	the	the	DET
ap-990	181	48	“	"	PUNCT
ap-990	181	49	aachen	aachen	PROPN
ap-990	181	50	smartchair	smartchair	NOUN
ap-990	181	51	”	"	PUNCT
ap-990	181	52	68	68	NUM
ap-990	181	53	a.	a.	NOUN
ap-990	181	54	aleksandrowicz	aleksandrowicz	NOUN
ap-990	181	55	,	,	PUNCT
ap-990	181	56	s.	s.	PROPN
ap-990	181	57	leonhardt	leonhardt	PROPN
ap-990	181	58	identification	identification	NOUN
ap-990	181	59	of	of	ADP
ap-990	181	60	nonlinear	nonlinear	ADJ
ap-990	181	61	systems	system	NOUN
ap-990	181	62	:	:	PUNCT
ap-990	181	63	volterra	volterra	PROPN
ap-990	181	64	series	series	PROPN
ap-990	181	65	simplification	simplification	NOUN
ap-990	181	66	72	72	NUM
ap-990	181	67	a.	a.	NOUN
ap-990	181	68	novák	novák	NOUN
ap-990	181	69	inductive	inductive	PROPN
ap-990	181	70	contactless	contactless	PROPN
ap-990	181	71	distance	distance	NOUN
ap-990	181	72	measurement	measurement	NOUN
ap-990	181	73	intended	intend	VERB
ap-990	181	74	for	for	ADP
ap-990	181	75	a	a	DET
ap-990	181	76	gastric	gastric	ADJ
ap-990	181	77	electrical	electrical	ADJ
ap-990	181	78	implant	implant	NOUN
ap-990	181	79	76	76	NUM
ap-990	181	80	j.	j.	PROPN
ap-990	181	81	tomek	tomek	PROPN
ap-990	181	82	analysis	analysis	NOUN
ap-990	181	83	of	of	ADP
ap-990	181	84	mismatched	mismatched	ADJ
ap-990	181	85	first	first	ADJ
ap-990	181	86	order	order	NOUN
ap-990	181	87	a	a	DET
ap-990	181	88	priori	priori	ADJ
ap-990	181	89	information	information	NOUN
ap-990	181	90	in	in	ADP
ap-990	181	91	iterative	iterative	ADJ
ap-990	181	92	source	source	NOUN
ap-990	181	93	-	-	PUNCT
ap-990	181	94	channel	channel	NOUN
ap-990	181	95	decoding	decode	VERB
ap-990	181	96	80	80	NUM
ap-990	181	97	b.	b.	NOUN
ap-990	181	98	schotsch	schotsch	NOUN
ap-990	181	99	,	,	PUNCT
ap-990	181	100	p.	p.	NOUN
ap-990	181	101	vary	vary	NOUN
ap-990	181	102	,	,	PUNCT
ap-990	181	103	t.	t.	PROPN
ap-990	181	104	clevorn	clevorn	ADJ
ap-990	181	105	automated	automate	VERB
ap-990	181	106	classification	classification	NOUN
ap-990	181	107	of	of	ADP
ap-990	181	108	analysisand	analysisand	NOUN
ap-990	181	109	reference	reference	NOUN
ap-990	181	110	cells	cell	NOUN
ap-990	181	111	for	for	ADP
ap-990	181	112	cancer	cancer	NOUN
ap-990	181	113	diagnostics	diagnostic	NOUN
ap-990	181	114	in	in	ADP
ap-990	181	115	microscopic	microscopic	ADJ
ap-990	181	116	images	image	NOUN
ap-990	181	117	of	of	ADP
ap-990	181	118	epithelial	epithelial	ADJ
ap-990	181	119	cells	cell	NOUN
ap-990	181	120	from	from	ADP
ap-990	181	121	the	the	DET
ap-990	181	122	oral	oral	ADJ
ap-990	181	123	mucosa	mucosa	NOUN
ap-990	181	124	86	86	NUM
ap-990	181	125	t.	t.	PROPN
ap-990	181	126	e.	e.	PROPN
ap-990	181	127	schneider	schneider	PROPN
ap-990	181	128	stick	stick	PROPN
ap-990	181	129	based	base	VERB
ap-990	181	130	speckle	speckle	NOUN
ap-990	181	131	reduction	reduction	NOUN
ap-990	181	132	for	for	ADP
ap-990	181	133	real	real	ADJ
ap-990	181	134	-	-	PUNCT
ap-990	181	135	time	time	NOUN
ap-990	181	136	processing	processing	NOUN
ap-990	181	137	of	of	ADP
ap-990	181	138	oct	oct	NOUN
ap-990	181	139	images	image	NOUN
ap-990	181	140	on	on	ADP
ap-990	181	141	an	an	DET
ap-990	181	142	fpga	fpga	PROPN
ap-990	181	143	91	91	PROPN
ap-990	181	144	h.	h.	NOUN
ap-990	181	145	luecken	luecken	VERB
ap-990	181	146	,	,	PUNCT
ap-990	181	147	g.	g.	PROPN
ap-990	181	148	tech	tech	PROPN
ap-990	181	149	,	,	PUNCT
ap-990	181	150	r.	r.	PROPN
ap-990	181	151	schwann	schwann	PROPN
ap-990	181	152	,	,	PUNCT
ap-990	181	153	g.	g.	PROPN
ap-990	181	154	kappen	kappen	VERB
ap-990	181	155	the	the	DET
ap-990	181	156	impact	impact	NOUN
ap-990	181	157	of	of	ADP
ap-990	181	158	connecting	connect	VERB
ap-990	181	159	distributed	distribute	VERB
ap-990	181	160	generation	generation	NOUN
ap-990	181	161	to	to	ADP
ap-990	181	162	the	the	DET
ap-990	181	163	distribution	distribution	NOUN
ap-990	181	164	system	system	NOUN
ap-990	181	165	96	96	NUM
ap-990	181	166	e.	e.	PROPN
ap-990	181	167	v.	v.	PROPN
ap-990	181	168	mgaya	mgaya	PROPN
ap-990	181	169	,	,	PUNCT
ap-990	181	170	z.	z.	PROPN
ap-990	181	171	müller	müller	PROPN
ap-990	181	172	improved	improve	VERB
ap-990	181	173	evaluation	evaluation	NOUN
ap-990	181	174	of	of	ADP
ap-990	181	175	planar	planar	ADJ
ap-990	181	176	calibration	calibration	NOUN
ap-990	181	177	standards	standard	NOUN
ap-990	181	178	using	use	VERB
ap-990	181	179	the	the	DET
ap-990	181	180	tdr	tdr	PROPN
ap-990	181	181	preselection	preselection	NOUN
ap-990	181	182	method	method	VERB
ap-990	181	183	102	102	NUM
ap-990	181	184	j.	j.	PROPN
ap-990	181	185	vancl	vancl	PROPN
ap-990	181	186	speaker	speaker	NOUN
ap-990	181	187	non	non	ADJ
ap-990	181	188	-	-	ADJ
ap-990	181	189	speech	speech	ADJ
ap-990	181	190	event	event	NOUN
ap-990	181	191	recognition	recognition	NOUN
ap-990	181	192	with	with	ADP
ap-990	181	193	standard	standard	ADJ
ap-990	181	194	speech	speech	NOUN
ap-990	181	195	datasets	dataset	NOUN
ap-990	181	196	107	107	NUM
ap-990	181	197	j.	j.	PROPN
ap-990	181	198	rajnoha	rajnoha	PROPN